6.19.7-rc1 review
Subsystems (10)
Sasha Levin
maintainerPATCH 139/311 net: phy: register phy led_triggers during probe to avoid AB-BA deadlock
In reply to Sasha LevinFrom: Andrew Lunn <[email protected]>
commit c8dbdc6e380e7e96a51706db3e4b7870d8a9402d upstream.
There is an AB-BA deadlock when both LEDS_TRIGGER_NETDEV and LED_TRIGGER_PHY are enabled:
[ 1362.049207] [<8054e4b8>] led_trigger_register+0x5c/0x1fc <-- Trying to get lock "triggers_list_lock" via down_write(&triggers_list_lock); [ 1362.054536] [<80662830>] phy_led_triggers_register+0xd0/0x234 [ 1362.060329] [<8065e200>] phy_attach_direct+0x33c/0x40c [ 1362.065489] [<80651fc4>] phylink_fwnode_phy_connect+0x15c/0x23c [ 1362.071480] [<8066ee18>] mtk_open+0x7c/0xba0 [ 1362.075849] [<806d714c>] __dev_open+0x280/0x2b0 [ 1362.080384] [<806d7668>] __dev_change_flags+0x244/0x24c [ 1362.085598] [<806d7698>] dev_change_flags+0x28/0x78 [ 1362.090528] [<807150e4>] dev_ioctl+0x4c0/0x654 <-- Hold lock "rtnl_mutex" by calling rtnl_lock(); [ 1362.094985] [<80694360>] sock_ioctl+0x2f4/0x4e0 [ 1362.099567] [<802e9c4c>] sys_ioctl+0x32c/0xd8c [ 1362.104022] [<80014504>] syscall_common+0x34/0x58
Here LED_TRIGGER_PHY is registering LED triggers during phy_attach while holding RTNL and then taking triggers_list_lock.
[ 1362.191101] [<806c2640>] register_netdevice_notifier+0x60/0x168 <-- Trying to get lock "rtnl_mutex" via rtnl_lock(); [ 1362.197073] [<805504ac>] netdev_trig_activate+0x194/0x1e4 [ 1362.202490] [<8054e28c>] led_trigger_set+0x1d4/0x360 <-- Hold lock "triggers_list_lock" by down_read(&triggers_list_lock); [ 1362.207511] [<8054eb38>] led_trigger_write+0xd8/0x14c [ 1362.212566] [<80381d98>] sysfs_kf_bin_write+0x80/0xbc [ 1362.217688] [<8037fcd8>] kernfs_fop_write_iter+0x17c/0x28c [ 1362.223174] [<802cbd70>] vfs_write+0x21c/0x3c4 [ 1362.227712] [<802cc0c4>] ksys_write+0x78/0x12c [ 1362.232164] [<80014504>] syscall_common+0x34/0x58
Here LEDS_TRIGGER_NETDEV is being enabled on an LED. It first takes triggers_list_lock and then RTNL. A classical AB-BA deadlock.
phy_led_triggers_registers() does not require the RTNL, it does not make any calls into the network stack which require protection. There is also no requirement the PHY has been attached to a MAC, the triggers only make use of phydev state. This allows the call to phy_led_triggers_registers() to be placed elsewhere. PHY probe() and release() don't hold RTNL, so solving the AB-BA deadlock.
--- drivers/net/phy/phy_device.c | 25 +++++++++++++++++-------- 1 file changed, 17 insertions(+), 8 deletions(-)
drivers/net/phy/phy_device.c +17 -8
@@ -1763,8 +1763,6 @@ int phy_attach_direct(struct net_device *dev, struct phy_device *phydev, goto error;phy_resume (phydev); -if (!phydev->is_on_sfp_module) -phy_led_triggers_register (phydev);/** * If the external phy used by current mac interface is managed by
@@ -1879,9 +1877,6 @@ void phy_detach(struct phy_device *phydev) phydev->phy_link_change =NULL ; phydev->phylink =NULL ; -if (!phydev->is_on_sfp_module) -phy_led_triggers_unregister (phydev); -if (phydev->mdio.dev.driver)module_put (phydev->mdio.dev.driver->owner);
@@ -3512,16 +3507,27 @@ static int phy_probe(struct device *dev)/* Set the state to READY by default */ phydev->state =PHY_READY ; +/* Register the PHY LED triggers */ +if (!phydev->is_on_sfp_module) +phy_led_triggers_register (phydev); +/* Get the LEDs from the device tree, and instantiate standard * LEDs for them. */ -if (IS_ENABLED (CONFIG_PHYLIB_LEDS ) && !phy_driver_is_genphy (phydev)) +if (IS_ENABLED (CONFIG_PHYLIB_LEDS ) && !phy_driver_is_genphy (phydev)) { err =of_phy_leds (phydev); +if (err) + goto out; + } + +return 0 ; out: +if (!phydev->is_on_sfp_module) +phy_led_triggers_unregister (phydev); +/* Re-assert the reset signal on error */ -if (err) -phy_device_reset (phydev,1 ); +phy_device_reset (phydev,1 );return err; }
@@ -3535,6 +3541,9 @@ static int phy_remove(struct device *dev)if (IS_ENABLED (CONFIG_PHYLIB_LEDS ) && !phy_driver_is_genphy (phydev))phy_leds_unregister (phydev); +if (!phydev->is_on_sfp_module) +phy_led_triggers_unregister (phydev); + phydev->state =PHY_DOWN ;sfp_bus_del_upstream (phydev->sfp_bus);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 140/311 IB/mthca: Add missed mthca_unmap_user_db() for mthca_create_srq()
In reply to Sasha LevinFrom: Jason Gunthorpe <[email protected]>
commit 117942ca43e2e3c3d121faae530989931b7f67e1 upstream.
Fix a user triggerable leak on the system call failure path.
--- drivers/infiniband/hw/mthca/mthca_provider.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/infiniband/hw/mthca/mthca_provider.c +3 -2
@@ -428,6 +428,8 @@ static int mthca_create_srq(struct ib_srq *ibsrq,if (context &&ib_copy_to_udata (udata, &srq->srqn,sizeof (__u32))) {mthca_free_srq (to_mdev (ibsrq->device), srq); +mthca_unmap_user_db (to_mdev (ibsrq->device), &context->uar, + context->db_tab, ucmd.db_index);return -EFAULT ; }
@@ -436,6 +438,7 @@ static int mthca_create_srq(struct ib_srq *ibsrq,static int mthca_destroy_srq (struct ib_srq *srq,struct ib_udata *udata) { +mthca_free_srq (to_mdev (srq->device),to_msrq (srq));if (udata) {struct mthca_ucontext *context = rdma_udata_to_drv_context(
@@ -446,8 +449,6 @@ static int mthca_destroy_srq(struct ib_srq *srq, struct ib_udata *udata)mthca_unmap_user_db (to_mdev (srq->device), &context->uar, context->db_tab,to_msrq (srq)->db_index); } - -mthca_free_srq (to_mdev (srq->device),to_msrq (srq));return 0 ; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 141/311 RDMA/irdma: Fix kernel stack leak in irdma_create_user_ah()
In reply to Sasha LevinFrom: Jason Gunthorpe <[email protected]>
commit 74586c6da9ea222a61c98394f2fc0a604748438c upstream.
struct irdma_create_ah_resp { // 8 bytes, no padding
__u32 ah_id; // offset 0 - SET (uresp.ah_id = ah->sc_ah.ah_info.ah_idx)
__u8 rsvd[4]; // offset 4 - NEVER SET <- LEAK};
rsvd[4]: 4 bytes of stack memory leaked unconditionally. Only ah_id is assigned before ib_respond_udata().
The reserved members of the structure were not zeroed.
--- drivers/infiniband/hw/irdma/verbs.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/infiniband/hw/irdma/verbs.c +1 -1
@@ -5209,7 +5209,7 @@ static int irdma_create_user_ah(struct ib_ah *ibah,#define IRDMA_CREATE_AH_MIN_RESP_LEN offsetofend(struct irdma_create_ah_resp, rsvd) struct irdma_ah *ah = container_of(ibah, struct irdma_ah, ibah);struct irdma_device *iwdev =to_iwdev (ibah->pd->device); -struct irdma_create_ah_resp uresp; +struct irdma_create_ah_resp uresp = {};struct irdma_ah *parent_ah;int err;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 142/311 RDMA/ionic: Fix kernel stack leak in ionic_create_cq()
In reply to Sasha LevinFrom: Jason Gunthorpe <[email protected]>
commit faa72102b178c7ae6c6afea23879e7c84fc59b4e upstream.
struct ionic_cq_resp resp {
__u32 cqid[2]; // offset 0 - PARTIALLY SET (see below)
__u8 udma_mask; // offset 8 - SET (resp.udma_mask = vcq->udma_mask)
__u8 rsvd[7]; // offset 9 - NEVER SET <- LEAK};
rsvd[7]: 7 bytes of stack memory leaked unconditionally.
cqid[2]: The loop at line 1256 iterates over udma_idx but skips indices where !(vcq->udma_mask & BIT(udma_idx)). The array has 2 entries but udma_count could be 1, meaning cqid[1] might never be written via ionic_create_cq_common(). If udma_mask only has bit 0 set, cqid[1] (4 bytes) is also leaked. So potentially 11 bytes leaked.
--- drivers/infiniband/hw/ionic/ionic_controlpath.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/infiniband/hw/ionic/ionic_controlpath.c +1 -1
@@ -1218,7 +1218,7 @@ int ionic_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr, rdma_udata_to_drv_context(udata, struct ionic_ctx, ibctx);struct ionic_vcq *vcq =to_ionic_vcq (ibcq);struct ionic_tbl_buf buf = {}; -struct ionic_cq_resp resp; +struct ionic_cq_resp resp = {};struct ionic_cq_req req;int udma_idx =0 , rc;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 143/311 ksmbd: Compare MACs in constant time
In reply to Sasha LevinFrom: Eric Biggers <[email protected]>
commit c5794709bc9105935dbedef8b9cf9c06f2b559fa upstream.
To prevent timing attacks, MAC comparisons need to be constant-time. Replace the memcmp() with the correct function, crypto_memneq().
--- fs/smb/server/Kconfig | 1 + fs/smb/server/auth.c | 4 +++- fs/smb/server/smb2pdu.c | 5 +++-- 3 files changed, 7 insertions(+), 3 deletions(-)
fs/smb/server/Kconfig +1 -0
@@ -13,6 +13,7 @@ config SMB_SERVER select CRYPTO_LIB_MD5 select CRYPTO_LIB_SHA256 select CRYPTO_LIB_SHA512 + select CRYPTO_LIB_UTILS select CRYPTO_CMAC select CRYPTO_AEAD2 select CRYPTO_CCM
fs/smb/server/auth.c +3 -1
@@ -15,6 +15,7 @@#include <crypto/aead.h> #include <crypto/md5.h> #include <crypto/sha2.h> +#include <crypto/utils.h> #include <linux/random.h> #include <linux/scatterlist.h>
@@ -165,7 +166,8 @@ int ksmbd_auth_ntlmv2(struct ksmbd_conn *conn, struct ksmbd_session *sess, ntlmv2_rsp,CIFS_HMAC_MD5_HASH_SIZE , sess->sess_key); -if (memcmp (ntlmv2->ntlmv2_hash, ntlmv2_rsp,CIFS_HMAC_MD5_HASH_SIZE ) !=0 ) +if (crypto_memneq (ntlmv2->ntlmv2_hash, ntlmv2_rsp, +CIFS_HMAC_MD5_HASH_SIZE ))return -EINVAL ;return 0 ; }
fs/smb/server/smb2pdu.c +3 -2
@@ -4,6 +4,7 @@* Copyright (C) 2018 Samsung Electronics Co., Ltd. */ +#include <crypto/utils.h> #include <linux/inetdevice.h> #include <net/addrconf.h> #include <linux/syscalls.h>
@@ -8879,7 +8880,7 @@ int smb2_check_sign_req(struct ksmbd_work *work)ksmbd_sign_smb2_pdu (work->conn, work->sess->sess_key, iov,1 , signature); -if (memcmp (signature, signature_req,SMB2_SIGNATURE_SIZE )) { +if (crypto_memneq (signature, signature_req,SMB2_SIGNATURE_SIZE )) {pr_err ("bad smb2 signature\n" );return 0 ; }
@@ -8967,7 +8968,7 @@ int smb3_check_sign_req(struct ksmbd_work *work)if (ksmbd_sign_smb3_pdu (conn, signing_key, iov,1 , signature))return 0 ; -if (memcmp (signature, signature_req,SMB2_SIGNATURE_SIZE )) { +if (crypto_memneq (signature, signature_req,SMB2_SIGNATURE_SIZE )) {pr_err ("bad smb2 signature\n" );return 0 ; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 144/311 cpufreq: intel_pstate: Fix crash during turbo disable
In reply to Sasha LevinFrom: Srinivas Pandruvada <[email protected]>
commit 6b050482ec40569429d963ac52afa878691b04c9 upstream.
When the system is booted with kernel command line argument "nosmt" or "maxcpus" to limit the number of CPUs, disabling turbo via:
echo 1 > /sys/devices/system/cpu/intel_pstate/no_turbo
results in a crash:
PF: supervisor read access in kernel mode
PF: error_code(0x0000) - not-present page
PGD 0 P4D 0
Oops: Oops: 0000 [#1] SMP PTI ... RIP: 0010:store_no_turbo+0x100/0x1f0 ...
This occurs because for_each_possible_cpu() returns CPUs even if they are not online. For those CPUs, all_cpu_data[] will be NULL. Since commit 973207ae3d7c ("cpufreq: intel_pstate: Rearrange max frequency updates handling code"), all_cpu_data[] is dereferenced even for CPUs which are not online, causing the NULL pointer dereference.
To fix that, pass CPU number to intel_pstate_update_max_freq() and use all_cpu_data[] for those CPUs for which there is a valid cpufreq policy.
--- drivers/cpufreq/intel_pstate.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-)
drivers/cpufreq/intel_pstate.c +5 -5
@@ -1476,13 +1476,13 @@ static void __intel_pstate_update_max_freq(struct cpufreq_policy *policy,refresh_frequency_limits (policy); } -static bool intel_pstate_update_max_freq (struct cpudata *cpudata) +static bool intel_pstate_update_max_freq (int cpu) { - struct cpufreq_policy *policy __free(put_cpufreq_policy) = cpufreq_cpu_get(cpudata->cpu); + struct cpufreq_policy *policy __free(put_cpufreq_policy) = cpufreq_cpu_get(cpu);if (!policy)return false; -__intel_pstate_update_max_freq (policy, cpudata); +__intel_pstate_update_max_freq (policy, all_cpu_data[cpu]);return true; }
@@ -1501,7 +1501,7 @@ static void intel_pstate_update_limits_for_all(void)int cpu;for_each_possible_cpu (cpu) -intel_pstate_update_max_freq (all_cpu_data[cpu]); +intel_pstate_update_max_freq (cpu);mutex_lock (&hybrid_capacity_lock);
@@ -1908,7 +1908,7 @@ static void intel_pstate_notify_work(struct work_struct *work)struct cpudata *cpudata = container_of(to_delayed_work(work), struct cpudata, hwp_notify_work); -if (intel_pstate_update_max_freq (cpudata)) { +if (intel_pstate_update_max_freq (cpudata->cpu)) {/* * The driver will not be unregistered while this function is * running, so update the capacity without acquiring the driver
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 145/311 arm64: gcs: Do not set PTE_SHARED on GCS mappings if FEAT_LPA2 is enabled
In reply to Sasha LevinFrom: Catalin Marinas <[email protected]>
commit 8a85b3131225a8c8143ba2ae29c0eef8c1f9117f upstream.
When FEAT_LPA2 is enabled, bits 8-9 of the PTE replace the shareability attribute with bits 50-51 of the output address. The _PAGE_GCS{,_RO} definitions include the PTE_SHARED bits as 0b11 (this matches the other _PAGE_* definitions) but using this macro directly leads to the following panic when enabling GCS on a system/model with LPA2:
Unable to handle kernel paging request at virtual address fffff1ffc32d8008
Mem abort info:
ESR = 0x0000000096000004
EC = 0x25: DABT (current EL), IL = 32 bits
SET = 0, FnV = 0
EA = 0, S1PTW = 0
FSC = 0x04: level 0 translation fault
Data abort info:
ISV = 0, ISS = 0x00000004, ISS2 = 0x00000000
CM = 0, WnR = 0, TnD = 0, TagAccess = 0
GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
swapper pgtable: 4k pages, 52-bit VAs, pgdp=0000000060f4d000
[fffff1ffc32d8008] pgd=100000006184b003, p4d=0000000000000000
Internal error: Oops: 0000000096000004 [#1] SMP
CPU: 0 UID: 0 PID: 513 Comm: gcs_write_fault Tainted: G M 7.0.0-rc1 #1 PREEMPT
Tainted: [M]=MACHINE_CHECK
Hardware name: QEMU QEMU Virtual Machine, BIOS 2025.02-8+deb13u1 11/08/2025
pstate: 03402005 (nzcv daif +PAN -UAO +TCO +DIT -SSBS BTYPE=--)
pc : zap_huge_pmd+0x168/0x468
lr : zap_huge_pmd+0x2c/0x468
sp : ffff800080beb660
x29: ffff800080beb660 x28: fff00000c2058180 x27: ffff800080beb898
x26: fff00000c2058180 x25: ffff800080beb820 x24: 00c800010b600f41
x23: ffffc1ffc30af1a8 x22: fff00000c2058180 x21: 0000ffff8dc00000
x20: fff00000c2bc6370 x19: ffff800080beb898 x18: ffff800080bebb60
x17: 0000000000000000 x16: 0000000000000000 x15: 0000000000000007
x14: 000000000000000a x13: 0000aaaacbbbffff x12: 0000000000000000
x11: 0000ffff8ddfffff x10: 00000000000001fe x9 : 0000ffff8ddfffff
x8 : 0000ffff8de00000 x7 : 0000ffff8da00000 x6 : fff00000c2bc6370
x5 : 0000ffff8da00000 x4 : 000000010b600000 x3 : ffffc1ffc0000000
x2 : fff00000c2058180 x1 : fffff1ffc32d8000 x0 : 000000c00010b600
Call trace:
zap_huge_pmd+0x168/0x468 (P)
unmap_page_range+0xd70/0x1560
unmap_single_vma+0x48/0x80
unmap_vmas+0x90/0x180
unmap_region+0x88/0xe4
vms_complete_munmap_vmas+0xf8/0x1e0
do_vmi_align_munmap+0x158/0x180
do_vmi_munmap+0xac/0x160
__vm_munmap+0xb0/0x138
vm_munmap+0x14/0x20
gcs_free+0x70/0x80
mm_release+0x1c/0xc8
exit_mm_release+0x28/0x38
do_exit+0x190/0x8ec
do_group_exit+0x34/0x90
get_signal+0x794/0x858
arch_do_signal_or_restart+0x11c/0x3e0
exit_to_user_mode_loop+0x10c/0x17c
el0_da+0x8c/0x9c
el0t_64_sync_handler+0xd0/0xf0
el0t_64_sync+0x198/0x19c
Code: aa1603e2 d34cfc00 cb813001 8b011861 (f9400420)Similarly to how the kernel handles protection_map[], use a gcs_page_prot variable to store the protection bits and clear PTE_SHARED if LPA2 is enabled.
Also remove the unused PAGE_GCS{,_RO} macros.
--- arch/arm64/include/asm/pgtable-prot.h | 3 --- arch/arm64/mm/mmap.c | 8 ++++++-- 2 files changed, 6 insertions(+), 5 deletions(-)
arch/arm64/include/asm/pgtable-prot.h +0 -3
@@ -164,9 +164,6 @@ static inline bool __pure lpa2_is_enabled(void)#define _PAGE_GCS (_PAGE_DEFAULT | PTE_NG | PTE_UXN | PTE_WRITE | PTE_USER)#define _PAGE_GCS_RO (_PAGE_DEFAULT | PTE_NG | PTE_UXN | PTE_USER) -#define PAGE_GCS __pgprot(_PAGE_GCS) -#define PAGE_GCS_RO __pgprot(_PAGE_GCS_RO) -#define PIE_E0 ( \ PIRx_ELx_PERM_PREP(pte_pi_index(_PAGE_GCS), PIE_GCS) | \ PIRx_ELx_PERM_PREP(pte_pi_index(_PAGE_GCS_RO), PIE_R) | \
arch/arm64/mm/mmap.c +6 -2
@@ -34,6 +34,8 @@ static pgprot_t protection_map[16] __ro_after_init = { [VM_SHARED | VM_EXEC | VM_WRITE | VM_READ] = PAGE_SHARED_EXEC }; +static ptdesc_t gcs_page_prot__ro_after_init = _PAGE_GCS_RO; +/* * You really shouldn't be using read() or write() on /dev/mem. This might go * away in the future.
@@ -73,9 +75,11 @@ static int __init adjust_protection_map(void) protection_map[VM_EXEC |VM_SHARED ] =PAGE_EXECONLY ; } -if (lpa2_is_enabled ()) +if (lpa2_is_enabled ()) {for (int i =0 ; i <ARRAY_SIZE (protection_map); i++)pgprot_val (protection_map[i]) &= ~PTE_SHARED ; + gcs_page_prot &= ~PTE_SHARED ; + }return 0 ; }
@@ -87,7 +91,7 @@ pgprot_t vm_get_page_prot(vm_flags_t vm_flags)/* Short circuit GCS to avoid bloating the table. */ if (system_supports_gcs () && (vm_flags &VM_SHADOW_STACK )) { - prot = _PAGE_GCS_RO; + prot = gcs_page_prot; }else { prot =pgprot_val (protection_map[vm_flags & (VM_READ |VM_WRITE |VM_EXEC |VM_SHARED )]);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 146/311 net/sched: ets: fix divide by zero in the offload path
In reply to Sasha LevinFrom: Davide Caratti <[email protected]>
commit e35626f610f3d2b7953ccddf6a77453da22b3a9e upstream.
Offloading ETS requires computing each class' WRR weight: this is done by averaging over the sums of quanta as 'q_sum' and 'q_psum'. Using unsigned int, the same integer size as the individual DRR quanta, can overflow and even cause division by zero, like it happened in the following splat:
Oops: divide error: 0000 [#1] SMP PTI CPU: 13 UID: 0 PID: 487 Comm: tc Tainted: G E 6.19.0-virtme #45 PREEMPT(full) Tainted: [E]=UNSIGNED_MODULE Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011 RIP: 0010:ets_offload_change+0x11f/0x290 [sch_ets] Code: e4 45 31 ff eb 03 41 89 c7 41 89 cb 89 ce 83 f9 0f 0f 87 b7 00 00 00 45 8b 08 31 c0 45 01 cc 45 85 c9 74 09 41 6b c4 64 31 d2 <41> f7 f2 89 c2 44 29 fa 45 89 df 41 83 fb 0f 0f 87 c7 00 00 00 44 RSP: 0018:ffffd0a180d77588 EFLAGS: 00010246 RAX: 00000000ffffff38 RBX: ffff8d3d482ca000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffffd0a180d77660 RBP: ffffd0a180d77690 R08: ffff8d3d482ca2d8 R09: 00000000fffffffe R10: 0000000000000000 R11: 0000000000000000 R12: 00000000fffffffe R13: ffff8d3d472f2000 R14: 0000000000000003 R15: 0000000000000000 FS: 00007f440b6c2740(0000) GS:ffff8d3dc9803000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 000000003cdd2000 CR3: 0000000007b58002 CR4: 0000000000172ef0 Call Trace: <TASK> ets_qdisc_change+0x870/0xf40 [sch_ets] qdisc_create+0x12b/0x540 tc_modify_qdisc+0x6d7/0xbd0 rtnetlink_rcv_msg+0x168/0x6b0 netlink_rcv_skb+0x5c/0x110 netlink_unicast+0x1d6/0x2b0 netlink_sendmsg+0x22e/0x470 ____sys_sendmsg+0x38a/0x3c0 ___sys_sendmsg+0x99/0xe0 __sys_sendmsg+0x8a/0xf0 do_syscall_64+0x111/0xf80 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7f440b81c77e Code: 4d 89 d8 e8 d4 bc 00 00 4c 8b 5d f8 41 8b 93 08 03 00 00 59 5e 48 83 f8 fc 74 11 c9 c3 0f 1f 80 00 00 00 00 48 8b 45 10 0f 05 <c9> c3 83 e2 39 83 fa 08 75 e7 e8 13 ff ff ff 0f 1f 00 f3 0f 1e fa RSP: 002b:00007fff951e4c10 EFLAGS: 00000202 ORIG_RAX: 000000000000002e RAX: ffffffffffffffda RBX: 0000000000481820 RCX: 00007f440b81c77e RDX: 0000000000000000 RSI: 00007fff951e4cd0 RDI: 0000000000000003 RBP: 00007fff951e4c20 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000202 R12: 00007fff951f4fa8 R13: 00000000699ddede R14: 00007f440bb01000 R15: 0000000000486980 </TASK> Modules linked in: sch_ets(E) netdevsim(E) ---[ end trace 0000000000000000 ]--- RIP: 0010:ets_offload_change+0x11f/0x290 [sch_ets] Code: e4 45 31 ff eb 03 41 89 c7 41 89 cb 89 ce 83 f9 0f 0f 87 b7 00 00 00 45 8b 08 31 c0 45 01 cc 45 85 c9 74 09 41 6b c4 64 31 d2 <41> f7 f2 89 c2 44 29 fa 45 89 df 41 83 fb 0f 0f 87 c7 00 00 00 44 RSP: 0018:ffffd0a180d77588 EFLAGS: 00010246 RAX: 00000000ffffff38 RBX: ffff8d3d482ca000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffffd0a180d77660 RBP: ffffd0a180d77690 R08: ffff8d3d482ca2d8 R09: 00000000fffffffe R10: 0000000000000000 R11: 0000000000000000 R12: 00000000fffffffe R13: ffff8d3d472f2000 R14: 0000000000000003 R15: 0000000000000000 FS: 00007f440b6c2740(0000) GS:ffff8d3dc9803000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 000000003cdd2000 CR3: 0000000007b58002 CR4: 0000000000172ef0 Kernel panic - not syncing: Fatal exception Kernel Offset: 0x30000000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff) ---[ end Kernel panic - not syncing: Fatal exception ]---
Fix this using 64-bit integers for 'q_sum' and 'q_psum'.
--- net/sched/sch_ets.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-)
net/sched/sch_ets.c +8 -4
@@ -115,12 +115,12 @@ static void ets_offload_change(struct Qdisc *sch)struct ets_sched *q =qdisc_priv (sch);struct tc_ets_qopt_offload qopt;unsigned int w_psum_prev =0 ; -unsigned int q_psum =0 ; -unsigned int q_sum =0 ;unsigned int quantum;unsigned int w_psum;unsigned int weight;unsigned int i; +u64 q_psum =0 ; +u64 q_sum =0 ;if (!tc_can_offload (dev) || !dev->netdev_ops->ndo_setup_tc)return ;
@@ -138,8 +138,12 @@ static void ets_offload_change(struct Qdisc *sch)for (i =0 ; i < q->nbands; i++) { quantum = q->classes[i].quantum; - q_psum += quantum; - w_psum = quantum ? q_psum *100 / q_sum :0 ; +if (quantum) { + q_psum += quantum; + w_psum =div64_u64 (q_psum *100 , q_sum); + }else { + w_psum =0 ; + } weight = w_psum - w_psum_prev; w_psum_prev = w_psum;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 147/311 nfsd: Fix cred ref leak in nfsd_nl_threads_set_doit().
In reply to Sasha LevinFrom: Kuniyuki Iwashima <[email protected]>
commit 1cb968a2013ffa8112d52ebe605009ea1c6a582c upstream.
syzbot reported memory leak of struct cred. [0]
nfsd_nl_threads_set_doit() passes get_current_cred() to nfsd_svc(), but put_cred() is not called after that.
The cred is finally passed down to _svc_xprt_create(), which calls get_cred() with the cred for struct svc_xprt.
The ownership of the refcount by get_current_cred() is not transferred to anywhere and is just leaked.
nfsd_svc() is also called from write_threads(), but it does not bump file->f_cred there.
nfsd_nl_threads_set_doit() is called from sendmsg() and current->cred does not go away.
Let's use current_cred() in nfsd_nl_threads_set_doit().
[0]:
BUG: memory leak
unreferenced object 0xffff888108b89480 (size 184):
comm "syz-executor", pid 5994, jiffies 4294943386
hex dump (first 32 bytes):
01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
backtrace (crc 369454a7):
kmemleak_alloc_recursive include/linux/kmemleak.h:44 [inline]
slab_post_alloc_hook mm/slub.c:4958 [inline]
slab_alloc_node mm/slub.c:5263 [inline]
kmem_cache_alloc_noprof+0x412/0x580 mm/slub.c:5270
prepare_creds+0x22/0x600 kernel/cred.c:185
copy_creds+0x44/0x290 kernel/cred.c:286
copy_process+0x7a7/0x2870 kernel/fork.c:2086
kernel_clone+0xac/0x6e0 kernel/fork.c:2651
__do_sys_clone+0x7f/0xb0 kernel/fork.c:2792
do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
do_syscall_64+0xa4/0xf80 arch/x86/entry/syscall_64.c:94
entry_SYSCALL_64_after_hwframe+0x77/0x7f--- fs/nfsd/nfsctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
fs/nfsd/nfsctl.c +1 -1
@@ -1642,7 +1642,7 @@ int nfsd_nl_threads_set_doit(struct sk_buff *skb, struct genl_info *info) scope =nla_data (attr); } - ret =nfsd_svc (nrpools, nthreads, net,get_current_cred (), scope); + ret =nfsd_svc (nrpools, nthreads, net,current_cred (), scope);if (ret >0 ) ret =0 ; out_unlock:
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 148/311 tracing: Fix WARN_ON in tracing_buffers_mmap_close
In reply to Sasha LevinFrom: Qing Wang <[email protected]>
commit e39bb9e02b68942f8e9359d2a3efe7d37ae6be0e upstream.
When a process forks, the child process copies the parent's VMAs but the user_mapped reference count is not incremented. As a result, when both the parent and child processes exit, tracing_buffers_mmap_close() is called twice. On the second call, user_mapped is already 0, causing the function to return -ENODEV and triggering a WARN_ON.
Normally, this isn't an issue as the memory is mapped with VM_DONTCOPY set. But this is only a hint, and the application can call madvise(MADVISE_DOFORK) which resets the VM_DONTCOPY flag. When the application does that, it can trigger this issue on fork.
Fix it by incrementing the user_mapped reference count without re-mapping the pages in the VMA's open callback.
--- include/linux/ring_buffer.h | 1 + kernel/trace/ring_buffer.c | 21 +++++++++++++++++++++ kernel/trace/trace.c | 13 +++++++++++++ 3 files changed, 35 insertions(+)
include/linux/ring_buffer.h +1 -0
@@ -248,6 +248,7 @@ int trace_rb_cpu_prepare(unsigned int cpu, struct hlist_node *node);int ring_buffer_map (struct trace_buffer *buffer,int cpu,struct vm_area_struct *vma); +void ring_buffer_map_dup (struct trace_buffer *buffer,int cpu);int ring_buffer_unmap (struct trace_buffer *buffer,int cpu);int ring_buffer_map_get_reader (struct trace_buffer *buffer,int cpu);#endif /* _LINUX_RING_BUFFER_H */
kernel/trace/ring_buffer.c +21 -0
@@ -7292,6 +7292,27 @@ int ring_buffer_map(struct trace_buffer *buffer, int cpu,return err; } +/* +* This is called when a VMA is duplicated (e.g., on fork()) to increment +* the user_mapped counter without remapping pages. +*/ +void ring_buffer_map_dup (struct trace_buffer *buffer,int cpu) +{ +struct ring_buffer_per_cpu *cpu_buffer; + +if (WARN_ON (!cpumask_test_cpu (cpu, buffer->cpumask))) +return ; + + cpu_buffer = buffer->buffers[cpu]; + +guard (mutex)(&cpu_buffer->mapping_lock); + +if (cpu_buffer->user_mapped) +__rb_inc_dec_mapped (cpu_buffer, true); +else +WARN (1 ,"Unexpected buffer stat, it should be mapped" ); +} +int ring_buffer_unmap (struct trace_buffer *buffer,int cpu) {struct ring_buffer_per_cpu *cpu_buffer;
kernel/trace/trace.c +13 -0
@@ -8999,6 +8999,18 @@ static inline int get_snapshot_map(struct trace_array *tr) { return 0; }static inline void put_snapshot_map (struct trace_array *tr) { }#endif +/* +* This is called when a VMA is duplicated (e.g., on fork()) to increment +* the user_mapped counter without remapping pages. +*/ +static void tracing_buffers_mmap_open (struct vm_area_struct *vma) +{ +struct ftrace_buffer_info *info = vma->vm_file->private_data; +struct trace_iterator *iter = &info->iter; + +ring_buffer_map_dup (iter->array_buffer->buffer, iter->cpu_file); +} +static void tracing_buffers_mmap_close (struct vm_area_struct *vma) {struct ftrace_buffer_info *info = vma->vm_file->private_data;
@@ -9018,6 +9030,7 @@ static int tracing_buffers_may_split(struct vm_area_struct *vma, unsigned long a }static const struct vm_operations_struct tracing_buffers_vmops = { + .open = tracing_buffers_mmap_open, .close = tracing_buffers_mmap_close, .may_split = tracing_buffers_may_split, };
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 149/311 scsi: target: Fix recursive locking in __configfs_open_file()
In reply to Sasha LevinFrom: Prithvi Tambewagh <[email protected]>
commit 14d4ac19d1895397532eec407433c5d74d9da53b upstream.
In flush_write_buffer, &p->frag_sem is acquired and then the loaded store function is called, which, here, is target_core_item_dbroot_store(). This function called filp_open(), following which these functions were called (in reverse order), according to the call trace:
down_read
__configfs_open_file
do_dentry_open
vfs_open
do_open
path_openat
do_filp_open
file_open_name
filp_open
target_core_item_dbroot_store
flush_write_buffer
configfs_write_iter
target_core_item_dbroot_store() tries to validate the new file path by trying to open the file path provided to it; however, in this case, the bug report shows:
db_root: not a directory: /sys/kernel/config/target/dbroot
indicating that the same configfs file was tried to be opened, on which it is currently working on. Thus, it is trying to acquire frag_sem semaphore of the same file of which it already holds the semaphore obtained in flush_write_buffer(), leading to acquiring the semaphore in a nested manner and a possibility of recursive locking.
Fix this by modifying target_core_item_dbroot_store() to use kern_path() instead of filp_open() to avoid opening the file using filesystem-specific function __configfs_open_file(), and further modifying it to make this fix compatible.
--- drivers/target/target_core_configfs.c | 15 ++++++--------- 1 file changed, 6 insertions(+), 9 deletions(-)
drivers/target/target_core_configfs.c +6 -9
@@ -108,8 +108,8 @@ static ssize_t target_core_item_dbroot_store(struct config_item *item, const char *page, size_t count) { ssize_t read_bytes; -struct file *fp;ssize_t r = -EINVAL ; +struct path path = {};mutex_lock (&target_devices_lock);if (target_devices) {
@@ -131,17 +131,14 @@ static ssize_t target_core_item_dbroot_store(struct config_item *item, db_root_stage[read_bytes -1 ] ='\0' ;/* validate new db root before accepting it */ - fp =filp_open (db_root_stage,O_RDONLY ,0 ); -if (IS_ERR (fp)) { + r =kern_path (db_root_stage,LOOKUP_FOLLOW |LOOKUP_DIRECTORY , &path); +if (r) {pr_err ("db_root: cannot open: %s\n" , db_root_stage); +if (r == -ENOTDIR ) +pr_err ("db_root: not a directory: %s\n" , db_root_stage); goto unlock; } -if (!S_ISDIR (file_inode (fp)->i_mode)) { -filp_close (fp,NULL ); -pr_err ("db_root: not a directory: %s\n" , db_root_stage); - goto unlock; - } -filp_close (fp,NULL ); +path_put (&path);strscpy (db_root, db_root_stage);pr_debug ("Target_Core_ConfigFS: db_root set to %s\n" , db_root);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 150/311 mm: thp: deny THP for files on anonymous inodes
In reply to Sasha LevinFrom: Deepanshu Kartikey <[email protected]>
commit dd085fe9a8ebfc5d10314c60452db38d2b75e609 upstream.
file_thp_enabled() incorrectly allows THP for files on anonymous inodes (e.g. guest_memfd and secretmem). These files are created via alloc_file_pseudo(), which does not call get_write_access() and leaves inode->i_writecount at 0. Combined with S_ISREG(inode->i_mode) being true, they appear as read-only regular files when CONFIG_READ_ONLY_THP_FOR_FS is enabled, making them eligible for THP collapse.
Anonymous inodes can never pass the inode_is_open_for_write() check since their i_writecount is never incremented through the normal VFS open path. The right thing to do is to exclude them from THP eligibility altogether, since CONFIG_READ_ONLY_THP_FOR_FS was designed for real filesystem files (e.g. shared libraries), not for pseudo-filesystem inodes.
For guest_memfd, this allows khugepaged and MADV_COLLAPSE to create large folios in the page cache via the collapse path, but the guest_memfd fault handler does not support large folios. This triggers WARN_ON_ONCE(folio_test_large(folio)) in kvm_gmem_fault_user_mapping().
For secretmem, collapse_file() tries to copy page contents through the direct map, but secretmem pages are removed from the direct map. This can result in a kernel crash:
BUG: unable to handle page fault for address: ffff88810284d000
RIP: 0010:memcpy_orig+0x16/0x130
Call Trace:
collapse_file
hpage_collapse_scan_file
madvise_collapseSecretmem is not affected by the crash on upstream as the memory failure recovery handles the failed copy gracefully, but it still triggers confusing false memory failure reports:
Memory failure: 0x106d96f: recovery action for clean unevictable LRU page: Recovered
Check IS_ANON_FILE(inode) in file_thp_enabled() to deny THP for all anonymous inode files.
--- mm/huge_memory.c | 3 +++ 1 file changed, 3 insertions(+)
mm/huge_memory.c +3 -0
@@ -94,6 +94,9 @@ static inline bool file_thp_enabled(struct vm_area_struct *vma) inode =file_inode (vma->vm_file); +if (IS_ANON_FILE (inode)) +return false; +return !inode_is_open_for_write (inode) &&S_ISREG (inode->i_mode); }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 151/311 Squashfs: check metadata block offset is within range
In reply to Sasha LevinFrom: Phillip Lougher <[email protected]>
commit fdb24a820a5832ec4532273282cbd4f22c291a0d upstream.
Syzkaller reports a "general protection fault in squashfs_copy_data"
This is ultimately caused by a corrupted index look-up table, which produces a negative metadata block offset.
This is subsequently passed to squashfs_copy_data (via squashfs_read_metadata) where the negative offset causes an out of bounds access.
The fix is to check that the offset is within range in squashfs_read_metadata. This will trap this and other cases.
--- fs/squashfs/cache.c | 3 +++ 1 file changed, 3 insertions(+)
fs/squashfs/cache.c +3 -0
@@ -344,6 +344,9 @@ int squashfs_read_metadata(struct super_block *sb, void *buffer,if (unlikely (length <0 ))return -EIO ; +if (unlikely (*offset <0 || *offset >=SQUASHFS_METADATA_SIZE )) +return -EIO ; +while (length) { entry =squashfs_cache_get (sb, msblk->block_cache, *block,0 );if (entry->error) {
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 152/311 drbd: fix "LOGIC BUG" in drbd_al_begin_io_nonblock()
In reply to Sasha LevinFrom: Lars Ellenberg <[email protected]>
commit ab140365fb62c0bdab22b2f516aff563b2559e3b upstream.
Even though we check that we "should" be able to do lc_get_cumulative() while holding the device->al_lock spinlock, it may still fail, if some other code path decided to do lc_try_lock() with bad timing.
If that happened, we logged "LOGIC BUG for enr=...", but still did not return an error.
The rest of the code now assumed that this request has references for the relevant activity log extents.
The implcations are that during an active resync, mutual exclusivity of resync versus application IO is not guaranteed. And a potential crash at this point may not realizs that these extents could have been target of in-flight IO and would need to be resynced just in case.
Also, once the request completes, it will give up activity log references it does not even hold, which will trigger a BUG_ON(refcnt == 0) in lc_put().
Fix:
Do not crash the kernel for a condition that is harmless during normal
operation: also catch "e->refcnt == 0", not only "e == NULL" when being noisy about "al_complete_io() called on inactive extent %u\n".
And do not try to be smart and "guess" whether something will work, then be surprised when it does not.
Deal with the fact that it may or may not work. If it does not, remember a possible "partially in activity log" state (only possible for requests that cross extent boundaries), and return an error code from drbd_al_begin_io_nonblock().
A latter call for the same request will then resume from where we left off.
--- drivers/block/drbd/drbd_actlog.c | 53 +++++++++++++----------------- drivers/block/drbd/drbd_interval.h | 5 ++- 2 files changed, 27 insertions(+), 31 deletions(-)
drivers/block/drbd/drbd_actlog.c +23 -30
@@ -483,38 +483,20 @@ void drbd_al_begin_io(struct drbd_device *device, struct drbd_interval *i)int drbd_al_begin_io_nonblock (struct drbd_device *device,struct drbd_interval *i) { -struct lru_cache *al = device->act_log;/* for bios crossing activity log extent boundaries, * we may need to activate two extents in one go */ unsigned first = i->sector >> (AL_EXTENT_SHIFT -9 );unsigned last = i->size ==0 ? first : (i->sector + (i->size >>9 ) -1 ) >> (AL_EXTENT_SHIFT -9 ); -unsigned nr_al_extents; -unsigned available_update_slots;unsigned enr; -D_ASSERT (device, first <= last); - - nr_al_extents =1 + last - first;/* worst case: all touched extends are cold. */ - available_update_slots =min (al->nr_elements - al->used, - al->max_pending_changes - al->pending_changes); - -/* We want all necessary updates for a given request within the same transaction -* We could first check how many updates are *actually* needed, -* and use that instead of the worst-case nr_al_extents */ -if (available_update_slots < nr_al_extents) { -/* Too many activity log extents are currently "hot". -* -* If we have accumulated pending changes already, -* we made progress. -* -* If we cannot get even a single pending change through, -* stop the fast path until we made some progress, -* or requests to "cold" extents could be starved. */ -if (!al->pending_changes) -__set_bit (__LC_STARVING, &device->act_log->flags); -return -ENOBUFS ; +if (i->partially_in_al_next_enr) { +D_ASSERT (device, first < i->partially_in_al_next_enr); +D_ASSERT (device, last >= i->partially_in_al_next_enr); + first = i->partially_in_al_next_enr; } +D_ASSERT (device, first <= last); +/* Is resync active in this area? */ for (enr = first; enr <= last; enr++) {struct lc_element *tmp;
@@ -529,14 +511,21 @@ int drbd_al_begin_io_nonblock(struct drbd_device *device, struct drbd_interval * } } -/* Checkout the refcounts. -* Given that we checked for available elements and update slots above, -* this has to be successful. */ +/* Try to checkout the refcounts. */ for (enr = first; enr <= last; enr++) {struct lc_element *al_ext; al_ext =lc_get_cumulative (device->act_log, enr); -if (!al_ext) -drbd_info (device,"LOGIC BUG for enr=%u\n" , enr); + +if (!al_ext) { +/* Did not work. We may have exhausted the possible +* changes per transaction. Or raced with someone +* "locking" it against changes. +* Remember where to continue from. +*/ +if (enr > first) + i->partially_in_al_next_enr = enr; +return -ENOBUFS ; + } }return 0 ; }
@@ -556,7 +545,11 @@ void drbd_al_complete_io(struct drbd_device *device, struct drbd_interval *i)for (enr = first; enr <= last; enr++) { extent =lc_find (device->act_log, enr); -if (!extent) { +/* Yes, this masks a bug elsewhere. However, during normal +* operation this is harmless, so no need to crash the kernel +* by the BUG_ON(refcount == 0) in lc_put(). +*/ +if (!extent || extent->refcnt ==0 ) {drbd_err (device,"al_complete_io() called on inactive extent %u\n" , enr);continue ; }
drivers/block/drbd/drbd_interval.h +4 -1
@@ -8,12 +8,15 @@struct drbd_interval {struct rb_node rb;sector_t sector;/* start sector of the interval */ -unsigned int size;/* size in bytes */ sector_t end;/* highest interval end in subtree */ +unsigned int size;/* size in bytes */ unsigned int local:1 /* local or remote request? */ ;unsigned int waiting:1 ;/* someone is waiting for completion */ unsigned int completed:1 ;/* this has been completed already; * ignore for conflict detection */ + +/* to resume a partially successful drbd_al_begin_io_nonblock(); */ +unsigned int partially_in_al_next_enr; };static inline void drbd_clear_interval (struct drbd_interval *i)
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 153/311 drbd: fix null-pointer dereference on local read error
In reply to Sasha LevinFrom: Christoph Böhmwalder <[email protected]>
commit 0d195d3b205ca90db30d70d09d7bb6909aac178f upstream.
In drbd_request_endio(), READ_COMPLETED_WITH_ERROR is passed to __req_mod() with a NULL peer_device:
__req_mod(req, what, NULL, &m);
The READ_COMPLETED_WITH_ERROR handler then unconditionally passes this NULL peer_device to drbd_set_out_of_sync(), which dereferences it, causing a null-pointer dereference.
Fix this by obtaining the peer_device via first_peer_device(device), matching how drbd_req_destroy() handles the same situation.
--- drivers/block/drbd/drbd_req.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/block/drbd/drbd_req.c +2 -1
@@ -621,7 +621,8 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what,break ;case READ_COMPLETED_WITH_ERROR : -drbd_set_out_of_sync (peer_device, req->i.sector, req->i.size); +drbd_set_out_of_sync (first_peer_device (device), + req->i.sector, req->i.size);drbd_report_io_error (device, req);__drbd_chk_io_error (device,DRBD_READ_ERROR ); fallthrough;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 154/311 xfs: fix xfs_group release bug in xfs_dax_notify_dev_failure
In reply to Sasha LevinFrom: "Darrick J. Wong" <[email protected]>
commit eb8550fb75a875657dc29e3925a40244ec6b6bd6 upstream.
Chris Mason reports that his AI tools noticed that we were using xfs_perag_put and xfs_group_put to release the group reference returned by xfs_group_next_range. However, the iterator function returns an object with an active refcount, which means that we must use the correct function to release the active refcount, which is _rele.
--- fs/xfs/xfs_notify_failure.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
fs/xfs/xfs_notify_failure.c +2 -2
@@ -293,7 +293,7 @@ xfs_dax_notify_dev_failure( error =xfs_alloc_read_agf (pag, tp,0 , &agf_bp);if (error) { -xfs_perag_put (pag); +xfs_perag_rele (pag);break ; }
@@ -329,7 +329,7 @@ xfs_dax_notify_dev_failure(if (rtg)xfs_rtgroup_unlock (rtg,XFS_RTGLOCK_RMAP );if (error) { -xfs_group_put (xg); +xfs_group_rele (xg);break ; } }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 155/311 xfs: Fix error pointer dereference
In reply to Sasha LevinFrom: Ethan Tidmore <[email protected]>
commit cddfa648f1ab99e30e91455be19cd5ade26338c2 upstream.
The function try_lookup_noperm() can return an error pointer and is not checked for one.
Add checks for error pointer in xrep_adoption_check_dcache() and xrep_adoption_zap_dcache().
Detected by Smatch:fs/xfs/scrub/orphanage.c:449 xrep_adoption_check_dcache() error:
'd_child' dereferencing possible ERR_PTR()
fs/xfs/scrub/orphanage.c:485 xrep_adoption_zap_dcache() error:
'd_child' dereferencing possible ERR_PTR()
--- fs/xfs/scrub/orphanage.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
fs/xfs/scrub/orphanage.c +6 -1
@@ -442,6 +442,11 @@ xrep_adoption_check_dcache(return 0 ; d_child =try_lookup_noperm (&qname, d_orphanage); +if (IS_ERR (d_child)) { +dput (d_orphanage); +return PTR_ERR (d_child); + } +if (d_child) {trace_xrep_adoption_check_child (sc->mp, d_child);
@@ -479,7 +484,7 @@ xrep_adoption_zap_dcache(return ; d_child =try_lookup_noperm (&qname, d_orphanage); -while (d_child !=NULL ) { +while (!IS_ERR_OR_NULL (d_child)) {trace_xrep_adoption_invalidate_child (sc->mp, d_child);ASSERT (d_is_negative (d_child));
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 156/311 smb: client: fix cifs_pick_channel when channels are equally loaded
In reply to Sasha LevinFrom: Henrique Carvalho <[email protected]>
commit 663c28469d3274d6456f206a6671c91493d85ff1 upstream.
cifs_pick_channel uses (start % chan_count) when channels are equally loaded, but that can return a channel that failed the eligibility checks.
Drop the fallback and return the scan-selected channel instead. If none is eligible, keep the existing behavior of using the primary channel.
--- fs/smb/client/transport.c | 21 ++++++++++----------- 1 file changed, 10 insertions(+), 11 deletions(-)
fs/smb/client/transport.c +10 -11
@@ -808,16 +808,21 @@ cifs_cancelled_callback(struct TCP_Server_Info *server, struct mid_q_entry *mid) }/* -* Return a channel (master if none) of @ses that can be used to send -* regular requests. +* cifs_pick_channel - pick an eligible channel for network operations * -* If we are currently binding a new channel (negprot/sess.setup), -* return the new incomplete channel. +* @ses: session reference +* +* Select an eligible channel (not terminating and not marked as needing +* reconnect), preferring the least loaded one. If no eligible channel is +* found, fall back to the primary channel (index 0). +* +* Return: TCP_Server_Info pointer for the chosen channel, or NULL if @ses is +* NULL. */ struct TCP_Server_Info *cifs_pick_channel (struct cifs_ses *ses) {uint index =0 ; -unsigned int min_in_flight =UINT_MAX , max_in_flight =0 ; +unsigned int min_in_flight =UINT_MAX ;struct TCP_Server_Info *server =NULL ;int i, start, cur;
@@ -847,14 +852,8 @@ struct TCP_Server_Info *cifs_pick_channel(struct cifs_ses *ses) min_in_flight = server->in_flight; index = cur; } -if (server->in_flight > max_in_flight) - max_in_flight = server->in_flight; } -/* if all channels are equally loaded, fall back to round-robin */ -if (min_in_flight == max_in_flight) - index = (uint )start % ses->chan_count; - server = ses->chans[index].server;spin_unlock (&ses->chan_lock);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 157/311 smb: client: fix broken multichannel with krb5+signing
In reply to Sasha LevinFrom: Paulo Alcantara <[email protected]>
commit d9d1e319b39ea685ede59319002d567c159d23c3 upstream.
When mounting a share with 'multichannel,max_channels=n,sec=krb5i', the client was duplicating signing key for all secondary channels, thus making the server fail all commands sent from secondary channels due to bad signatures.
Every channel has its own signing key, so when establishing a new channel with krb5 auth, make sure to use the new session key as the derived key to generate channel's signing key in SMB2_auth_kerberos().
Repro:
$ mount.cifs //srv/share /mnt -o multichannel,max_channels=4,sec=krb5i $ sleep 5 $ umount /mnt $ dmesg ... CIFS: VFS: sign fail cmd 0x5 message id 0x2 CIFS: VFS: \\srv SMB signature verification returned error = -13 CIFS: VFS: sign fail cmd 0x5 message id 0x2 CIFS: VFS: \\srv SMB signature verification returned error = -13 CIFS: VFS: sign fail cmd 0x4 message id 0x2 CIFS: VFS: \\srv SMB signature verification returned error = -13
--- fs/smb/client/smb2pdu.c | 22 ++++++++++------------ 1 file changed, 10 insertions(+), 12 deletions(-)
fs/smb/client/smb2pdu.c +10 -12
@@ -1715,19 +1715,17 @@ SMB2_auth_kerberos(struct SMB2_sess_data *sess_data) is_binding = (ses->ses_status ==SES_GOOD );spin_unlock (&ses->ses_lock); -/* keep session key if binding */ -if (!is_binding) { -kfree_sensitive (ses->auth_key.response); - ses->auth_key.response =kmemdup (msg->data, msg->sesskey_len, -GFP_KERNEL ); -if (!ses->auth_key.response) { -cifs_dbg (VFS ,"Kerberos can't allocate (%u bytes) memory\n" , - msg->sesskey_len); - rc = -ENOMEM ; - goto out_put_spnego_key; - } - ses->auth_key.len = msg->sesskey_len; +kfree_sensitive (ses->auth_key.response); + ses->auth_key.response =kmemdup (msg->data, + msg->sesskey_len, +GFP_KERNEL ); +if (!ses->auth_key.response) { +cifs_dbg (VFS ,"%s: can't allocate (%u bytes) memory\n" , + __func__, msg->sesskey_len); + rc = -ENOMEM ; + goto out_put_spnego_key; } + ses->auth_key.len = msg->sesskey_len; sess_data->iov[1 ].iov_base = msg->data + msg->sesskey_len; sess_data->iov[1 ].iov_len = msg->secblob_len;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 158/311 smb: client: Don't log plaintext credentials in cifs_set_cifscreds
In reply to Sasha LevinFrom: Thorsten Blum <[email protected]>
commit 2f37dc436d4e61ff7ae0b0353cf91b8c10396e4d upstream.
When debug logging is enabled, cifs_set_cifscreds() logs the key payload and exposes the plaintext username and password. Remove the debug log to avoid exposing credentials.
--- fs/smb/client/connect.c | 1 - 1 file changed, 1 deletion(-)
fs/smb/client/connect.c +0 -1
@@ -2233,7 +2233,6 @@ cifs_set_cifscreds(struct smb3_fs_context *ctx, struct cifs_ses *ses)/* find first : in payload */ payload = upayload->data; delim =strnchr (payload, upayload->datalen,':' ); -cifs_dbg (FYI ,"payload=%s\n" , payload);if (!delim) {cifs_dbg (FYI ,"Unable to find ':' in payload (datalen=%d)\n" , upayload->datalen);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 159/311 smb: client: fix oops due to uninitialised var in smb2_unlink()
In reply to Sasha LevinFrom: Paulo Alcantara <[email protected]>
commit 048efe129a297256d3c2088cf8d79515ff5ec864 upstream.
If SMB2_open_init() or SMB2_close_init() fails (e.g. reconnect), the iovs set @rqst will be left uninitialised, hence calling SMB2_open_free(), SMB2_close_free() or smb2_set_related() on them will oops.
Fix this by initialising @close_iov and @open_iov before setting them in @rqst.
--- fs/smb/client/smb2inode.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
fs/smb/client/smb2inode.c +3 -1
@@ -1208,6 +1208,7 @@ smb2_unlink(const unsigned int xid, struct cifs_tcon *tcon, const char *name,memset (resp_buftype,0 ,sizeof (resp_buftype));memset (rsp_iov,0 ,sizeof (rsp_iov)); +memset (open_iov,0 ,sizeof (open_iov)); rqst[0 ].rq_iov = open_iov; rqst[0 ].rq_nvec =ARRAY_SIZE (open_iov);
@@ -1232,14 +1233,15 @@ smb2_unlink(const unsigned int xid, struct cifs_tcon *tcon, const char *name, creq = rqst[0 ].rq_iov[0 ].iov_base; creq->ShareAccess =FILE_SHARE_DELETE_LE ; +memset (&close_iov,0 ,sizeof (close_iov)); rqst[1 ].rq_iov = &close_iov; rqst[1 ].rq_nvec =1 ; rc =SMB2_close_init (tcon, server, &rqst[1 ],COMPOUND_FID ,COMPOUND_FID , false); -smb2_set_related (&rqst[1 ]);if (rc) goto err_free; +smb2_set_related (&rqst[1 ]);if (retries) {for (int i =0 ; i <ARRAY_SIZE (rqst); i++)
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 160/311 scsi: core: Fix refcount leak for tagset_refcnt
In reply to Sasha LevinFrom: Junxiao Bi <[email protected]>
commit 1ac22c8eae81366101597d48360718dff9b9d980 upstream.
This leak will cause a hang when tearing down the SCSI host. For example, iscsid hangs with the following call trace:
[130120.652718] scsi_alloc_sdev: Allocation failure during SCSI scanning, some SCSI devices might not be configured
PID: 2528 TASK: ffff9d0408974e00 CPU: 3 COMMAND: "iscsid"
#0 [ffffb5b9c134b9e0] __schedule at ffffffff860657d4
#1 [ffffb5b9c134ba28] schedule at ffffffff86065c6f
#2 [ffffb5b9c134ba40] schedule_timeout at ffffffff86069fb0
#3 [ffffb5b9c134bab0] __wait_for_common at ffffffff8606674f
#4 [ffffb5b9c134bb10] scsi_remove_host at ffffffff85bfe84b
#5 [ffffb5b9c134bb30] iscsi_sw_tcp_session_destroy at ffffffffc03031c4 [iscsi_tcp] #6 [ffffb5b9c134bb48] iscsi_if_recv_msg at ffffffffc0292692 [scsi_transport_iscsi] #7 [ffffb5b9c134bb98] iscsi_if_rx at ffffffffc02929c2 [scsi_transport_iscsi] #8 [ffffb5b9c134bbf0] netlink_unicast at ffffffff85e551d6
#9 [ffffb5b9c134bc38] netlink_sendmsg at ffffffff85e554ef
--- drivers/scsi/scsi_scan.c | 1 + 1 file changed, 1 insertion(+)
drivers/scsi/scsi_scan.c +1 -0
@@ -361,6 +361,7 @@ static struct scsi_device *scsi_alloc_sdev(struct scsi_target *starget,* since we use this queue depth most of times. */ if (scsi_realloc_sdev_budget_map (sdev, depth)) { +kref_put (&sdev->host->tagset_refcnt, scsi_mq_free_tags);put_device (&starget->dev);kfree (sdev); goto out;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 161/311 scsi: ufs: core: Fix RPMB region size detection for UFS 2.2
In reply to Sasha LevinFrom: Alexey Charkov <[email protected]>
commit 2e6b5cd6a4b37a95b78cf8c39a979b58c915c8ed upstream.
Older UFS spec devices (2.2 and earlier) do not expose per-region RPMB sizes, as only one RPMB region is supported. In such cases, the size of the single RPMB region can be deduced from the Logical Block Count and Logical Block Size fields in the RPMB Unit Descriptor.
Add a fallback mechanism to calculate the RPMB region size from these fields if the device implements an older spec, so that the RPMB driver can work with such devices - otherwise it silently skips the whole RPMB.
Section 14.1.4.6 (RPMB Unit Descriptor)
--- drivers/ufs/core/ufshcd.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+)
drivers/ufs/core/ufshcd.c +20 -0
@@ -23,6 +23,7 @@#include <linux/pm_opp.h> #include <linux/regulator/consumer.h> #include <linux/sched/clock.h> +#include <linux/sizes.h> #include <linux/iopoll.h> #include <scsi/scsi_cmnd.h> #include <scsi/scsi_dbg.h>
@@ -5237,6 +5238,25 @@ static void ufshcd_lu_init(struct ufs_hba *hba, struct scsi_device *sdev) hba->dev_info.rpmb_region_size[1 ] = desc_buf[RPMB_UNIT_DESC_PARAM_REGION1_SIZE ]; hba->dev_info.rpmb_region_size[2 ] = desc_buf[RPMB_UNIT_DESC_PARAM_REGION2_SIZE ]; hba->dev_info.rpmb_region_size[3 ] = desc_buf[RPMB_UNIT_DESC_PARAM_REGION3_SIZE ]; + +if (hba->dev_info.wspecversion <=0x0220 ) { +/* +* These older spec chips have only one RPMB region, +* sized between 128 kB minimum and 16 MB maximum. +* No per region size fields are provided (respective +* REGIONX_SIZE fields always contain zeros), so get +* it from the logical block count and size fields for +* compatibility +* +* (See JESD220C-2_2 Section 14.1.4.6 +* RPMB Unit Descriptor,* offset 13h, 4 bytes) +*/ + hba->dev_info.rpmb_region_size[0 ] = + (get_unaligned_be64 (desc_buf + +RPMB_UNIT_DESC_PARAM_LOGICAL_BLK_COUNT ) + << desc_buf[RPMB_UNIT_DESC_PARAM_LOGICAL_BLK_SIZE ]) + /SZ_128K ; + } }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 162/311 mptcp: pm: avoid sending RM_ADDR over same subflow
In reply to Sasha LevinFrom: "Matthieu Baerts (NGI0)" <[email protected]>
commit fb8d0bccb221080630efcd9660c9f9349e53cc9e upstream.
RM_ADDR are sent over an active subflow, the first one in the subflows list. There is then a high chance the initial subflow is picked. With the in-kernel PM, when an endpoint is removed, a RM_ADDR is sent, then linked subflows are closed. This is done for each active MPTCP connection.
MPTCP endpoints are likely removed because the attached network is no longer available or usable. In this case, it is better to avoid sending this RM_ADDR over the subflow that is going to be removed, but prefer sending it over another active and non stale subflow, if any.
This modification avoids situations where the other end is not notified when a subflow is no longer usable: typically when the endpoint linked to the initial subflow is removed, especially on the server side.
--- net/mptcp/pm.c | 55 +++++++++++++++++++++++++++++++++++++++----------- 1 file changed, 43 insertions(+), 12 deletions(-)
net/mptcp/pm.c +43 -12
@@ -212,9 +212,24 @@ void mptcp_pm_send_ack(struct mptcp_sock *msk,spin_lock_bh (&msk->pm.lock); } -void mptcp_pm_addr_send_ack (struct mptcp_sock *msk) +static bool subflow_in_rm_list (const struct mptcp_subflow_context *subflow, +const struct mptcp_rm_list *rm_list) +{ +u8 i, id =subflow_get_local_id (subflow); + +for (i =0 ; i < rm_list->nr; i++) { +if (rm_list->ids[i] == id) +return true; + } + +return false; +} + +static void +mptcp_pm_addr_send_ack_avoid_list (struct mptcp_sock *msk, +const struct mptcp_rm_list *rm_list) { -struct mptcp_subflow_context *subflow, *alt =NULL ; +struct mptcp_subflow_context *subflow, *stale =NULL , *same_id =NULL ;msk_owned_by_me (msk);lockdep_assert_held (&msk->pm.lock);
@@ -224,19 +239,35 @@ void mptcp_pm_addr_send_ack(struct mptcp_sock *msk)return ;mptcp_for_each_subflow (msk, subflow) { -if (__mptcp_subflow_active (subflow)) { -if (!subflow->stale) { -mptcp_pm_send_ack (msk, subflow, false, false); -return ; - } +if (!__mptcp_subflow_active (subflow)) +continue ; -if (!alt) - alt = subflow; +if (unlikely (subflow->stale)) { +if (!stale) + stale = subflow; + }else if (unlikely (rm_list && +subflow_in_rm_list (subflow, rm_list))) { +if (!same_id) + same_id = subflow; + }else { + goto send_ack; } } -if (alt) -mptcp_pm_send_ack (msk, alt, false, false); +if (same_id) + subflow = same_id; +else if (stale) + subflow = stale; +else +return ; + +send_ack: +mptcp_pm_send_ack (msk, subflow, false, false); +} + +void mptcp_pm_addr_send_ack (struct mptcp_sock *msk) +{ +mptcp_pm_addr_send_ack_avoid_list (msk,NULL ); } int mptcp_pm_mp_prio_send_ack(struct mptcp_sock *msk,
@@ -470,7 +501,7 @@ int mptcp_pm_remove_addr(struct mptcp_sock *msk, const struct mptcp_rm_list *rm_ msk->pm.rm_list_tx = *rm_list; rm_addr |=BIT (MPTCP_RM_ADDR_SIGNAL );WRITE_ONCE (msk->pm.addr_signal, rm_addr); -mptcp_pm_addr_send_ack (msk); +mptcp_pm_addr_send_ack_avoid_list (msk, rm_list);return 0 ; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 163/311 mptcp: pm: in-kernel: always mark signal+subflow endp as used
In reply to Sasha LevinFrom: "Matthieu Baerts (NGI0)" <[email protected]>
commit 579a752464a64cb5f9139102f0e6b90a1f595ceb upstream.
Syzkaller managed to find a combination of actions that was generating this warning:
msk->pm.local_addr_used == 0
WARNING: net/mptcp/pm_kernel.c:1071 at __mark_subflow_endp_available net/mptcp/pm_kernel.c:1071 [inline], CPU#1: syz.2.17/961 WARNING: net/mptcp/pm_kernel.c:1071 at mptcp_nl_remove_subflow_and_signal_addr net/mptcp/pm_kernel.c:1103 [inline], CPU#1: syz.2.17/961 WARNING: net/mptcp/pm_kernel.c:1071 at mptcp_pm_nl_del_addr_doit+0x81d/0x8f0 net/mptcp/pm_kernel.c:1210, CPU#1: syz.2.17/961 Modules linked in: CPU: 1 UID: 0 PID: 961 Comm: syz.2.17 Not tainted 6.19.0-08368-gfafda3b4b06b #22 PREEMPT(full) Hardware name: QEMU Ubuntu 25.10 PC v2 (i440FX + PIIX, + 10.1 machine, 1996), BIOS 1.17.0-debian-1.17.0-1build1 04/01/2014 RIP: 0010:__mark_subflow_endp_available net/mptcp/pm_kernel.c:1071 [inline] RIP: 0010:mptcp_nl_remove_subflow_and_signal_addr net/mptcp/pm_kernel.c:1103 [inline] RIP: 0010:mptcp_pm_nl_del_addr_doit+0x81d/0x8f0 net/mptcp/pm_kernel.c:1210 Code: 89 c5 e8 46 30 6f fe e9 21 fd ff ff 49 83 ed 80 e8 38 30 6f fe 4c 89 ef be 03 00 00 00 e8 db 49 df fe eb ac e8 24 30 6f fe 90 <0f> 0b 90 e9 1d ff ff ff e8 16 30 6f fe eb 05 e8 0f 30 6f fe e8 9a RSP: 0018:ffffc90001663880 EFLAGS: 00010293 RAX: ffffffff82de1a6c RBX: 0000000000000000 RCX: ffff88800722b500 RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000 RBP: ffff8880158b22d0 R08: 0000000000010425 R09: ffffffffffffffff R10: ffffffff82de18ba R11: 0000000000000000 R12: ffff88800641a640 R13: ffff8880158b1880 R14: ffff88801ec3c900 R15: ffff88800641a650 FS: 00005555722c3500(0000) GS:ffff8880f909d000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f66346e0f60 CR3: 000000001607c000 CR4: 0000000000350ef0 Call Trace: <TASK> genl_family_rcv_msg_doit+0x117/0x180 net/netlink/genetlink.c:1115 genl_family_rcv_msg net/netlink/genetlink.c:1195 [inline] genl_rcv_msg+0x3a8/0x3f0 net/netlink/genetlink.c:1210 netlink_rcv_skb+0x16d/0x240 net/netlink/af_netlink.c:2550 genl_rcv+0x28/0x40 net/netlink/genetlink.c:1219 netlink_unicast_kernel net/netlink/af_netlink.c:1318 [inline] netlink_unicast+0x3e9/0x4c0 net/netlink/af_netlink.c:1344 netlink_sendmsg+0x4aa/0x5b0 net/netlink/af_netlink.c:1894 sock_sendmsg_nosec net/socket.c:727 [inline] __sock_sendmsg+0xc9/0xf0 net/socket.c:742 ____sys_sendmsg+0x272/0x3b0 net/socket.c:2592 ___sys_sendmsg+0x2de/0x320 net/socket.c:2646 __sys_sendmsg net/socket.c:2678 [inline] __do_sys_sendmsg net/socket.c:2683 [inline] __se_sys_sendmsg net/socket.c:2681 [inline] __x64_sys_sendmsg+0x110/0x1a0 net/socket.c:2681 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline] do_syscall_64+0x143/0x440 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7f66346f826d Code: ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 e8 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007ffc83d8bdc8 EFLAGS: 00000246 ORIG_RAX: 000000000000002e RAX: ffffffffffffffda RBX: 00007f6634985fa0 RCX: 00007f66346f826d RDX: 00000000040000b0 RSI: 0000200000000740 RDI: 0000000000000007 RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00007f6634985fa8 R13: 00007f6634985fac R14: 0000000000000000 R15: 0000000000001770 </TASK>
The actions that caused that seem to be:
- Set the MPTCP subflows limit to 0
- Create an MPTCP endpoint with both the 'signal' and 'subflow' flags
- Create a new MPTCP connection from a different address: an ADD_ADDR linked to the MPTCP endpoint will be sent ('signal' flag), but no subflows is initiated ('subflow' flag)
- Remove the MPTCP endpoint
In this case, msk->pm.local_addr_used has been kept to 0 -- because no subflows have been created -- but the corresponding bit in msk->pm.id_avail_bitmap has been cleared when the ADD_ADDR has been sent. This later causes a splat when removing the MPTCP endpoint because msk->pm.local_addr_used has been kept to 0.
Now, if an endpoint has both the signal and subflow flags, but it is not possible to create subflows because of the limits or the c-flag case, then the local endpoint counter is still incremented: the endpoint is used at the end. This avoids issues later when removing the endpoint and calling __mark_subflow_endp_available(), which expects msk->pm.local_addr_used to have been previously incremented if the endpoint was marked as used according to msk->pm.id_avail_bitmap.
Note that signal_and_subflow variable is reset to false when the limits and the c-flag case allows subflows creation. Also, local_addr_used is only incremented for non ID0 subflows.
--- net/mptcp/pm_kernel.c | 9 +++++++++ 1 file changed, 9 insertions(+)
net/mptcp/pm_kernel.c +9 -0
@@ -418,6 +418,15 @@ static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk) } exit: +/* If an endpoint has both the signal and subflow flags, but it is not +* possible to create subflows -- the 'while' loop body above never +* executed -- then still mark the endp as used, which is somehow the +* case. This avoids issues later when removing the endpoint and calling +* __mark_subflow_endp_available(), which expects the increment here. +*/ +if (signal_and_subflow && local.addr.id != msk->mpc_endpoint_id) + msk->pm.local_addr_used++; +mptcp_pm_nl_check_work_pending (msk); }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 164/311 selftests: mptcp: more stable simult_flows tests
In reply to Sasha LevinFrom: Paolo Abeni <[email protected]>
commit 8c09412e584d9bcc0e71d758ec1008d1c8d1a326 upstream.
By default, the netem qdisc can keep up to 1000 packets under its belly to deal with the configured rate and delay. The simult flows test-case simulates very low speed links, to avoid problems due to slow CPUs and the TCP stack tend to transmit at a slightly higher rate than the (virtual) link constraints.
All the above causes a relatively large amount of packets being enqueued in the netem qdiscs - the longer the transfer, the longer the queue - producing increasingly high TCP RTT samples and consequently increasingly larger receive buffer size due to DRS.
When the receive buffer size becomes considerably larger than the needed size, the tests results can flake, i.e. because minimal inaccuracy in the pacing rate can lead to a single subflow usage towards the end of the connection for a considerable amount of data.
Address the issue explicitly setting netem limits suitable for the configured link speeds and unflake all the affected tests.
--- tools/testing/selftests/net/mptcp/simult_flows.sh | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)
tools/testing/selftests/net/mptcp/simult_flows.sh +7 -4
@@ -237,10 +237,13 @@ run_test() for dev in ns2eth1 ns2eth2; do tc -n $ns2 qdisc del dev $dev root >/dev/null 2>&1 done - tc -n $ns1 qdisc add dev ns1eth1 root netem rate ${rate1}mbit $delay1 - tc -n $ns1 qdisc add dev ns1eth2 root netem rate ${rate2}mbit $delay2 - tc -n $ns2 qdisc add dev ns2eth1 root netem rate ${rate1}mbit $delay1 - tc -n $ns2 qdisc add dev ns2eth2 root netem rate ${rate2}mbit $delay2 + + # keep the queued pkts number low, or the RTT estimator will see + # increasing latency over time. + tc -n $ns1 qdisc add dev ns1eth1 root netem rate ${rate1}mbit $delay1 limit 50 + tc -n $ns1 qdisc add dev ns1eth2 root netem rate ${rate2}mbit $delay2 limit 50 + tc -n $ns2 qdisc add dev ns2eth1 root netem rate ${rate1}mbit $delay1 limit 50 + tc -n $ns2 qdisc add dev ns2eth2 root netem rate ${rate2}mbit $delay2 limit 50 # time is measured in ms, account for transfer size, aggregated link speed # and header overhead (10%)
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 165/311 selftests: mptcp: join: check RM_ADDR not sent over same subflow
In reply to Sasha LevinFrom: "Matthieu Baerts (NGI0)" <[email protected]>
commit 560edd99b5f58b2d4bbe3c8e51e1eed68d887b0e upstream.
This validates the previous commit: RM_ADDR were sent over the first found active subflow which could be the same as the one being removed. It is more likely to loose this notification.
For this check, RM_ADDR are explicitly dropped when trying to send them over the initial subflow, when removing the endpoint attached to it. If it is dropped, the test will complain because some RM_ADDR have not been received.
Note that only the RM_ADDR are dropped, to allow the linked subflow to be quickly and cleanly closed. To only drop those RM_ADDR, a cBPF byte code is used. If the IPTables commands fail, that's OK, the tests will continue to pass, but not validate this part. This can be ignored:
another subtest fully depends on such command, and will be marked as skipped.
The 'Fixes' tag here below is the same as the one from the previous
commit: this patch here is not fixing anything wrong in the selftests, but it validates the previous fix for an issue introduced by this commit ID.
--- .../testing/selftests/net/mptcp/mptcp_join.sh | 36 +++++++++++++++++++ 1 file changed, 36 insertions(+)
tools/testing/selftests/net/mptcp/mptcp_join.sh +36 -0
@@ -104,6 +104,24 @@ CBPF_MPTCP_SUBOPTION_ADD_ADDR="14, 6 0 0 65535, 6 0 0 0" +# IPv4: TCP hdr of 48B, a first suboption of 12B (DACK8), the RM_ADDR suboption +# generated using "nfbpf_compile '(ip[32] & 0xf0) == 0xc0 && ip[53] == 0x0c && +# (ip[66] & 0xf0) == 0x40'" +CBPF_MPTCP_SUBOPTION_RM_ADDR="13, + 48 0 0 0, + 84 0 0 240, + 21 0 9 64, + 48 0 0 32, + 84 0 0 240, + 21 0 6 192, + 48 0 0 53, + 21 0 4 12, + 48 0 0 66, + 84 0 0 240, + 21 0 1 64, + 6 0 0 65535, + 6 0 0 0" + init_partial() { capout=$(mktemp)
@@ -4222,6 +4240,14 @@ endpoint_tests() chk_subflow_nr "after no reject" 3 chk_mptcp_info subflows 2 subflows 2 + # To make sure RM_ADDR are sent over a different subflow, but + # allow the rest to quickly and cleanly close the subflow + local ipt=1 + ip netns exec "${ns2}" ${iptables} -I OUTPUT -s "10.0.1.2" \ + -p tcp -m tcp --tcp-option 30 \ + -m bpf --bytecode \ + "$CBPF_MPTCP_SUBOPTION_RM_ADDR" \ + -j DROP || ipt=0 local i for i in $(seq 3); do pm_nl_del_endpoint $ns2 1 10.0.1.2
@@ -4234,6 +4260,7 @@ endpoint_tests() chk_subflow_nr "after re-add id 0 ($i)" 3 chk_mptcp_info subflows 3 subflows 3 done + [ ${ipt} = 1 ] && ip netns exec "${ns2}" ${iptables} -D OUTPUT 1 mptcp_lib_kill_group_wait $tests_pid
@@ -4293,11 +4320,20 @@ endpoint_tests() chk_mptcp_info subflows 2 subflows 2 chk_mptcp_info add_addr_signal 2 add_addr_accepted 2 + # To make sure RM_ADDR are sent over a different subflow, but + # allow the rest to quickly and cleanly close the subflow + local ipt=1 + ip netns exec "${ns1}" ${iptables} -I OUTPUT -s "10.0.1.1" \ + -p tcp -m tcp --tcp-option 30 \ + -m bpf --bytecode \ + "$CBPF_MPTCP_SUBOPTION_RM_ADDR" \ + -j DROP || ipt=0 pm_nl_del_endpoint $ns1 42 10.0.1.1 sleep 0.5 chk_subflow_nr "after delete ID 0" 2 chk_mptcp_info subflows 2 subflows 2 chk_mptcp_info add_addr_signal 2 add_addr_accepted 2 + [ ${ipt} = 1 ] && ip netns exec "${ns1}" ${iptables} -D OUTPUT 1 pm_nl_add_endpoint $ns1 10.0.1.1 id 99 flags signal wait_mpj $ns2
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 166/311 selftests: mptcp: join: check removing signal+subflow endp
In reply to Sasha LevinFrom: "Matthieu Baerts (NGI0)" <[email protected]>
commit 1777f349ff41b62dfe27454b69c27b0bc99ffca5 upstream.
This validates the previous commit: endpoints with both the signal and subflow flags should always be marked as used even if it was not possible to create new subflows due to the MPTCP PM limits.
For this test, an extra endpoint is created with both the signal and the subflow flags, and limits are set not to create extra subflows. In this case, an ADD_ADDR is sent, but no subflows are created. Still, the local endpoint is marked as used, and no warning is fired when removing the endpoint, after having sent a RM_ADDR.
The 'Fixes' tag here below is the same as the one from the previous
commit: this patch here is not fixing anything wrong in the selftests, but it validates the previous fix for an issue introduced by this commit ID.
--- tools/testing/selftests/net/mptcp/mptcp_join.sh | 13 +++++++++++++ 1 file changed, 13 insertions(+)
tools/testing/selftests/net/mptcp/mptcp_join.sh +13 -0
@@ -2637,6 +2637,19 @@ remove_tests() chk_rst_nr 0 0 fi + # signal+subflow with limits, remove + if reset "remove signal+subflow with limits"; then + pm_nl_set_limits $ns1 0 0 + pm_nl_add_endpoint $ns1 10.0.2.1 flags signal,subflow + pm_nl_set_limits $ns2 0 0 + addr_nr_ns1=-1 speed=slow \ + run_tests $ns1 $ns2 10.0.1.1 + chk_join_nr 0 0 0 + chk_add_nr 1 1 + chk_rm_nr 1 0 invert + chk_rst_nr 0 0 + fi + # addresses remove if reset "remove addresses"; then pm_nl_set_limits $ns1 3 3
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 167/311 kbuild: Split .modinfo out from ELF_DETAILS
In reply to Sasha LevinFrom: Nathan Chancellor <[email protected]>
commit 8678591b47469fe16357234efef9b260317b8be4 upstream.
Commit 3e86e4d74c04 ("kbuild: keep .modinfo section in vmlinux.unstripped") added .modinfo to ELF_DETAILS while removing it from COMMON_DISCARDS, as it was needed in vmlinux.unstripped and ELF_DETAILS was present in all architecture specific vmlinux linker scripts. While this shuffle is fine for vmlinux, ELF_DETAILS and COMMON_DISCARDS may be used by other linker scripts, such as the s390 and x86 compressed boot images, which may not expect to have a .modinfo section. In certain circumstances, this could result in a bootloader failing to load the compressed kernel [1].
Commit ddc6cbef3ef1 ("s390/boot/vmlinux.lds.S: Ensure bzImage ends with SecureBoot trailer") recently addressed this for the s390 bzImage but the same bug remains for arm, parisc, and x86. The presence of .modinfo in the x86 bzImage was the root cause of the issue worked around with commit d50f21091358 ("kbuild: align modinfo section for Secureboot Authenticode EDK2 compat"). misc.c in arch/x86/boot/compressed includes lib/decompress_unzstd.c, which in turn includes lib/xxhash.c and its MODULE_LICENSE / MODULE_DESCRIPTION macros due to the STATIC definition.
Split .modinfo out from ELF_DETAILS into its own macro and handle it in all vmlinux linker scripts. Discard .modinfo in the places where it was previously being discarded from being in COMMON_DISCARDS, as it has never been necessary in those uses.
--- arch/alpha/kernel/vmlinux.lds.S | 1 + arch/arc/kernel/vmlinux.lds.S | 1 + arch/arm/boot/compressed/vmlinux.lds.S | 1 + arch/arm/kernel/vmlinux-xip.lds.S | 1 + arch/arm/kernel/vmlinux.lds.S | 1 + arch/arm64/kernel/vmlinux.lds.S | 1 + arch/csky/kernel/vmlinux.lds.S | 1 + arch/hexagon/kernel/vmlinux.lds.S | 1 + arch/loongarch/kernel/vmlinux.lds.S | 1 + arch/m68k/kernel/vmlinux-nommu.lds | 1 + arch/m68k/kernel/vmlinux-std.lds | 1 + arch/m68k/kernel/vmlinux-sun3.lds | 1 + arch/mips/kernel/vmlinux.lds.S | 1 + arch/nios2/kernel/vmlinux.lds.S | 1 + arch/openrisc/kernel/vmlinux.lds.S | 1 + arch/parisc/boot/compressed/vmlinux.lds.S | 1 + arch/parisc/kernel/vmlinux.lds.S | 1 + arch/powerpc/kernel/vmlinux.lds.S | 1 + arch/riscv/kernel/vmlinux.lds.S | 1 + arch/s390/kernel/vmlinux.lds.S | 1 + arch/sh/kernel/vmlinux.lds.S | 1 + arch/sparc/kernel/vmlinux.lds.S | 1 + arch/um/kernel/dyn.lds.S | 1 + arch/um/kernel/uml.lds.S | 1 + arch/x86/boot/compressed/vmlinux.lds.S | 2 +- arch/x86/kernel/vmlinux.lds.S | 1 + include/asm-generic/vmlinux.lds.h | 4 +++- 27 files changed, 29 insertions(+), 2 deletions(-)
arch/alpha/kernel/vmlinux.lds.S +1 -0
@@ -71,6 +71,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS DISCARDS
arch/arc/kernel/vmlinux.lds.S +1 -0
@@ -123,6 +123,7 @@ SECTIONS _end = . ; STABS_DEBUG + MODINFO ELF_DETAILS DISCARDS
arch/arm/boot/compressed/vmlinux.lds.S +1 -0
@@ -21,6 +21,7 @@ SECTIONS COMMON_DISCARDS *(.ARM.exidx*) *(.ARM.extab*) + *(.modinfo) *(.note.*) *(.rel.*) *(.printk_index)
arch/arm/kernel/vmlinux-xip.lds.S +1 -0
@@ -154,6 +154,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ARM_DETAILS ARM_ASSERTS
arch/arm/kernel/vmlinux.lds.S +1 -0
@@ -153,6 +153,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ARM_DETAILS ARM_ASSERTS
arch/arm64/kernel/vmlinux.lds.S +1 -0
@@ -349,6 +349,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS HEAD_SYMBOLS
arch/csky/kernel/vmlinux.lds.S +1 -0
@@ -109,6 +109,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS DISCARDS
arch/hexagon/kernel/vmlinux.lds.S +1 -0
@@ -62,6 +62,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS .hexagon.attributes 0 : { *(.hexagon.attributes) }
arch/loongarch/kernel/vmlinux.lds.S +1 -0
@@ -147,6 +147,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS #ifdef CONFIG_EFI_STUB
arch/m68k/kernel/vmlinux-nommu.lds +1 -0
@@ -85,6 +85,7 @@ SECTIONS { _end = .; STABS_DEBUG + MODINFO ELF_DETAILS /* Sections to be discarded */
arch/m68k/kernel/vmlinux-std.lds +1 -0
@@ -58,6 +58,7 @@ SECTIONS _end = . ; STABS_DEBUG + MODINFO ELF_DETAILS /* Sections to be discarded */
arch/m68k/kernel/vmlinux-sun3.lds +1 -0
@@ -51,6 +51,7 @@ __init_begin = .; _end = . ; STABS_DEBUG + MODINFO ELF_DETAILS /* Sections to be discarded */
arch/mips/kernel/vmlinux.lds.S +1 -0
@@ -217,6 +217,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS /* These must appear regardless of . */
arch/nios2/kernel/vmlinux.lds.S +1 -0
@@ -57,6 +57,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS DISCARDS
arch/openrisc/kernel/vmlinux.lds.S +1 -0
@@ -101,6 +101,7 @@ SECTIONS /* Throw in the debugging sections */ STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS /* Sections to be discarded -- must be last */
arch/parisc/boot/compressed/vmlinux.lds.S +1 -0
@@ -90,6 +90,7 @@ SECTIONS /* Sections to be discarded */ DISCARDS /DISCARD/ : { + *(.modinfo) #ifdef CONFIG_64BIT /* temporary hack until binutils is fixed to not emit these * for static binaries
arch/parisc/kernel/vmlinux.lds.S +1 -0
@@ -165,6 +165,7 @@ SECTIONS _end = . ; STABS_DEBUG + MODINFO ELF_DETAILS .note 0 : { *(.note) }
arch/powerpc/kernel/vmlinux.lds.S +1 -0
@@ -397,6 +397,7 @@ SECTIONS _end = . ; DWARF_DEBUG + MODINFO ELF_DETAILS DISCARDS
arch/riscv/kernel/vmlinux.lds.S +1 -0
@@ -170,6 +170,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS .riscv.attributes 0 : { *(.riscv.attributes) }
arch/s390/kernel/vmlinux.lds.S +1 -0
@@ -221,6 +221,7 @@ SECTIONS /* Debugging sections. */ STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS /*
arch/sh/kernel/vmlinux.lds.S +1 -0
@@ -89,6 +89,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS DISCARDS
arch/sparc/kernel/vmlinux.lds.S +1 -0
@@ -191,6 +191,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS DISCARDS
arch/um/kernel/dyn.lds.S +1 -0
@@ -172,6 +172,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS DISCARDS
arch/um/kernel/uml.lds.S +1 -0
@@ -113,6 +113,7 @@ SECTIONS STABS_DEBUG DWARF_DEBUG + MODINFO ELF_DETAILS DISCARDS
arch/x86/boot/compressed/vmlinux.lds.S +1 -1
@@ -88,7 +88,7 @@ SECTIONS /DISCARD/ : { *(.dynamic) *(.dynsym) *(.dynstr) *(.dynbss) *(.hash) *(.gnu.hash) - *(.note.*) + *(.note.*) *(.modinfo) } .got.plt (INFO) : {
arch/x86/kernel/vmlinux.lds.S +1 -0
@@ -424,6 +424,7 @@ SECTIONS .llvm_bb_addr_map : { *(.llvm_bb_addr_map) } #endif + MODINFO ELF_DETAILS DISCARDS
include/asm-generic/vmlinux.lds.h +3 -1
@@ -848,12 +848,14 @@/* Required sections not related to debugging. */ #define ELF_DETAILS \ - .modinfo : { *(.modinfo) . = ALIGN(8); } \ .comment 0 : { *(.comment) } \ .symtab 0 : { *(.symtab) } \ .strtab 0 : { *(.strtab) } \ .shstrtab 0 : { *(.shstrtab) } +#define MODINFO \ + .modinfo : { *(.modinfo) . = ALIGN(8); } +#ifdef CONFIG_GENERIC_BUG #define BUG_TABLE \ . = ALIGN(8); \
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 168/311 kbuild: Leave objtool binary around with 'make clean'
In reply to Sasha LevinFrom: Nathan Chancellor <[email protected]>
commit fdb12c8a24a453bdd6759979b6ef1e04ebd4beb4 upstream.
The difference between 'make clean' and 'make mrproper' is documented in 'make help' as:
clean - Remove most generated files but keep the config and
enough build support to build external modules
mrproper - Remove all generated files + config + various backup files
After commit 68b4fe32d737 ("kbuild: Add objtool to top-level clean target"), running 'make clean' then attempting to build an external module with the resulting build directory fails with
$ make ARCH=x86_64 O=build clean
$ make -C build M=... MO=... ... /bin/sh: line 1: .../build/tools/objtool/objtool: No such file or directory
as 'make clean' removes the objtool binary.
Split the objtool clean target into mrproper and clean like Kbuild does and remove all generated artifacts with 'make clean' except for the objtool binary, which is removed with 'make mrproper'. To avoid a small race when running the objtool clean target through both objtool_mrproper and objtool_clean when running 'make mrproper', modify objtool's clean up find command to avoid using find's '-delete' command by piping the files into 'xargs rm -f' like the rest of Kbuild does.
--- Makefile | 8 ++++---- tools/objtool/Makefile | 8 +++++--- 2 files changed, 9 insertions(+), 7 deletions(-)
Makefile +4 -4
@@ -1474,13 +1474,13 @@ ifneq ($(wildcard $(resolve_btfids_O)),) $(Q)$(MAKE) -sC $(srctree)/tools/bpf/resolve_btfids O=$(resolve_btfids_O) clean endif -PHONY += objtool_clean +PHONY += objtool_clean objtool_mrproper objtool_O = $(abspath $(objtree))/tools/objtool -objtool_clean: +objtool_clean objtool_mrproper: ifneq ($(wildcard $(objtool_O)),) - $(Q)$(MAKE) -sC $(abs_srctree)/tools/objtool O=$(objtool_O) srctree=$(abs_srctree) clean + $(Q)$(MAKE) -sC $(abs_srctree)/tools/objtool O=$(objtool_O) srctree=$(abs_srctree) $(patsubst objtool_%,%,$@) endif tools/: FORCE
@@ -1657,7 +1657,7 @@ PHONY += $(mrproper-dirs) mrproper $(mrproper-dirs): $(Q)$(MAKE) $(clean)=$(patsubst _mrproper_%,%,$@) -mrproper: clean $(mrproper-dirs) +mrproper: clean objtool_mrproper $(mrproper-dirs) $(call cmd,rmfiles) @find . $(RCS_FIND_IGNORE) \ \( -name '*.rmeta' \) \
tools/objtool/Makefile +5 -3
@@ -142,13 +142,15 @@ $(LIBSUBCMD)-clean: $(Q)$(RM) -r -- $(LIBSUBCMD_OUTPUT) clean: $(LIBSUBCMD)-clean - $(call QUIET_CLEAN, objtool) $(RM) $(OBJTOOL) - $(Q)find $(OUTPUT) -name '*.o' -delete -o -name '\.*.cmd' -delete -o -name '\.*.d' -delete + $(Q)find $(OUTPUT) \( -name '*.o' -o -name '\.*.cmd' -o -name '\.*.d' \) -type f -print | xargs $(RM) $(Q)$(RM) $(OUTPUT)arch/x86/lib/cpu-feature-names.c $(OUTPUT)fixdep $(Q)$(RM) $(OUTPUT)arch/x86/lib/inat-tables.c $(OUTPUT)fixdep $(Q)$(RM) -- $(OUTPUT)FEATURE-DUMP.objtool $(Q)$(RM) -r -- $(OUTPUT)feature +mrproper: clean + $(call QUIET_CLEAN, objtool) $(RM) $(OBJTOOL) + FORCE: -.PHONY: clean FORCE +.PHONY: clean mrproper FORCE
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 169/311 ASoC: sdca: Fix missing regmap dependencies in Kconfig
In reply to Sasha LevinFrom: Boris Faure <[email protected]>
commit bbb758a6943e19c483ab752cf8220140b46cf22c upstream.
The SDCA modules failed to build with modpost errors:
ERROR: modpost: "__devm_regmap_init_sdw" [sound/soc/sdca/snd-soc-sdca-class.ko] undefined!
ERROR: modpost: "__devm_regmap_init_sdw_mbq" [sound/soc/sdca/snd-soc-sdca-class-function.ko] undefined!
The issue occurs because:
- sdca_class.c calls devm_regmap_init_sdw() which requires REGMAP_SOUNDWIRE
- sdca_class_function.c calls devm_regmap_init_sdw_mbq_cfg() which requires REGMAP_SOUNDWIRE_MBQ
However, the Kconfig didn't select these dependencies, causing the symbols to be unavailable when the SDCA modules are built.
Fix this by adding:
- select REGMAP_SOUNDWIRE to SND_SOC_SDCA_CLASS
- select REGMAP_SOUNDWIRE_MBQ to SND_SOC_SDCA_CLASS_FUNCTION
This ensures the required regmap drivers are enabled when building SDCA support.
Configuration after fix:
CONFIG_SND_SOC_SDCA_CLASS=m
CONFIG_SND_SOC_SDCA_CLASS_FUNCTION=m
CONFIG_REGMAP_SOUNDWIRE=m
CONFIG_REGMAP_SOUNDWIRE_MBQ=m
--- sound/soc/sdca/Kconfig | 2 ++ 1 file changed, 2 insertions(+)
sound/soc/sdca/Kconfig +2 -0
@@ -46,12 +46,14 @@ config SND_SOC_SDCA_CLASS select SND_SOC_SDCA_FDL select SND_SOC_SDCA_HID select SND_SOC_SDCA_IRQ + select REGMAP_SOUNDWIRE help This option enables support for the SDCA Class driver which should support any class compliant SDCA part. config SND_SOC_SDCA_CLASS_FUNCTION tristate + select REGMAP_SOUNDWIRE_MBQ help This option enables support for the SDCA Class Function drivers, these implement the individual functions of the SDCA Class driver.
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 170/311 Revert "netfilter: nft_set_rbtree: validate open interval overlap"
In reply to Sasha LevinFrom: Greg Kroah-Hartman <[email protected]>
This reverts commit 6db2be971e3d70c9e3f85d39eff7103c2ee2f579 which is commit 648946966a08e4cb1a71619e3d1b12bd7642de7b upstream.
It is causing netfilter issues, so revert it for now.
--- include/net/netfilter/nf_tables.h | 4 -- net/netfilter/nf_tables_api.c | 21 ++------- net/netfilter/nft_set_rbtree.c | 71 +++++-------------------------- 3 files changed, 14 insertions(+), 82 deletions(-)
include/net/netfilter/nf_tables.h +0 -4
@@ -278,8 +278,6 @@ struct nft_userdata {unsigned char data[]; }; -#define NFT_SET_ELEM_INTERNAL_LAST 0x1 -/* placeholder structure for opaque set element backend representation. */ struct nft_elem_priv { };
@@ -289,7 +287,6 @@ struct nft_elem_priv { };* @key: element key * @key_end: closing element key * @data: element data -* @flags: flags * @priv: element private data and extensions */ struct nft_set_elem {
@@ -305,7 +302,6 @@ struct nft_set_elem {u32 buf[NFT_DATA_VALUE_MAXLEN /sizeof (u32)];struct nft_data val; } data; -u32 flags;struct nft_elem_priv *priv; };
net/netfilter/nf_tables_api.c +4 -17
@@ -7153,8 +7153,7 @@ static u32 nft_set_maxsize(const struct nft_set *set) }static int nft_add_set_elem (struct nft_ctx *ctx,struct nft_set *set, -const struct nlattr *attr,u32 nlmsg_flags, -bool last) +const struct nlattr *attr,u32 nlmsg_flags) {struct nft_expr *expr_array[NFT_SET_EXPR_MAX ] = {};struct nlattr *nla[NFTA_SET_ELEM_MAX +1 ];
@@ -7440,11 +7439,6 @@ static int nft_add_set_elem(struct nft_ctx *ctx, struct nft_set *set,if (flags) *nft_set_ext_flags (ext) = flags; -if (last) - elem.flags =NFT_SET_ELEM_INTERNAL_LAST ; -else - elem.flags =0 ; -if (obj) *nft_set_ext_obj (ext) = obj;
@@ -7608,8 +7602,7 @@ static int nf_tables_newsetelem(struct sk_buff *skb,nft_ctx_init (&ctx, net, skb, info->nlh, family, table,NULL , nla);nla_for_each_nested (attr, nla[NFTA_SET_ELEM_LIST_ELEMENTS ], rem) { - err =nft_add_set_elem (&ctx, set, attr, info->nlh->nlmsg_flags, -nla_is_last (attr, rem)); + err =nft_add_set_elem (&ctx, set, attr, info->nlh->nlmsg_flags);if (err <0 ) {NL_SET_BAD_ATTR (extack, attr);return err;
@@ -7733,7 +7726,7 @@ static void nft_trans_elems_destroy_abort(const struct nft_ctx *ctx, }static int nft_del_setelem (struct nft_ctx *ctx,struct nft_set *set, -const struct nlattr *attr,bool last) +const struct nlattr *attr) {struct nlattr *nla[NFTA_SET_ELEM_MAX +1 ];struct nft_set_ext_tmpl tmpl;
@@ -7801,11 +7794,6 @@ static int nft_del_setelem(struct nft_ctx *ctx, struct nft_set *set,if (flags) *nft_set_ext_flags (ext) = flags; -if (last) - elem.flags =NFT_SET_ELEM_INTERNAL_LAST ; -else - elem.flags =0 ; - trans =nft_trans_elem_alloc (ctx,NFT_MSG_DELSETELEM , set);if (trans ==NULL ) goto fail_trans;
@@ -7953,8 +7941,7 @@ static int nf_tables_delsetelem(struct sk_buff *skb,return nft_set_flush (&ctx, set, genmask);nla_for_each_nested (attr, nla[NFTA_SET_ELEM_LIST_ELEMENTS ], rem) { - err =nft_del_setelem (&ctx, set, attr, -nla_is_last (attr, rem)); + err =nft_del_setelem (&ctx, set, attr);if (err == -ENOENT &&NFNL_MSG_TYPE (info->nlh->nlmsg_type) ==NFT_MSG_DESTROYSETELEM )continue ;
net/netfilter/nft_set_rbtree.c +10 -61
136 lines not on this page · raw message
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 171/311 ARM: clean up the memset64() C wrapper
In reply to Sasha LevinFrom: Thomas Weißschuh <[email protected]>
commit b52343d1cb47bb27ca32a3f4952cc2fd3cd165bf upstream.
The current logic to split the 64-bit argument into its 32-bit halves is byte-order specific and a bit clunky. Use a union instead which is easier to read and works in all cases.
GCC still generates the same machine code.
While at it, rename the arguments of the __memset64() prototype to actually reflect their semantics.
--- arch/arm/include/asm/string.h | 14 +++++++++----- 1 file changed, 9 insertions(+), 5 deletions(-)
arch/arm/include/asm/string.h +9 -5
@@ -39,13 +39,17 @@ static inline void *memset32(uint32_t *p, uint32_t v, __kernel_size_t n) }#define __HAVE_ARCH_MEMSET64 -extern void *__memset64 (uint64_t *,uint32_t low,__kernel_size_t ,uint32_t hi); +extern void *__memset64 (uint64_t *,uint32_t first,__kernel_size_t ,uint32_t second);static inline void *memset64 (uint64_t *p,uint64_t v,__kernel_size_t n) { -if (IS_ENABLED (CONFIG_CPU_LITTLE_ENDIAN )) -return __memset64 (p, v, n *8 , v >>32 ); -else -return __memset64 (p, v >>32 , n *8 , v); +union { +uint64_t val; +struct { +uint32_t first, second; + }; + } word = { .val = v }; + +return __memset64 (p, word.first, n *8 , word.second); }/*
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 172/311 platform/x86: hp-bioscfg: Support allocations of larger data
In reply to Sasha LevinFrom: Mario Limonciello <[email protected]>
commit 916727cfdb72cd01fef3fa6746e648f8cb70e713 upstream.
Some systems have much larger amounts of enumeration attributes than have been previously encountered. This can lead to page allocation failures when using kcalloc(). Switch over to using kvcalloc() to allow larger allocations.
--- drivers/platform/x86/hp/hp-bioscfg/enum-attributes.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)
drivers/platform/x86/hp/hp-bioscfg/enum-attributes.c +6 -3
@@ -94,8 +94,11 @@ int hp_alloc_enumeration_data(void) bioscfg_drv.enumeration_instances_count =hp_get_instance_count (HP_WMI_BIOS_ENUMERATION_GUID ); - bioscfg_drv.enumeration_data =kcalloc (bioscfg_drv.enumeration_instances_count, -sizeof (*bioscfg_drv.enumeration_data),GFP_KERNEL ); +if (!bioscfg_drv.enumeration_instances_count) +return -EINVAL ; + bioscfg_drv.enumeration_data =kvcalloc (bioscfg_drv.enumeration_instances_count, +sizeof (*bioscfg_drv.enumeration_data),GFP_KERNEL ); +if (!bioscfg_drv.enumeration_data) { bioscfg_drv.enumeration_instances_count =0 ;return -ENOMEM ;
@@ -444,6 +447,6 @@ void hp_exit_enumeration_attributes(void) } bioscfg_drv.enumeration_instances_count =0 ; -kfree (bioscfg_drv.enumeration_data); +kvfree (bioscfg_drv.enumeration_data); bioscfg_drv.enumeration_data =NULL ; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 173/311 Bluetooth: Fix CIS host feature condition
In reply to Sasha LevinFrom: Mariusz Skamra <[email protected]>
commit 7cff9a40c6b0f72ccefdaf0ffe03cfac30348f51 upstream.
This fixes the condition for sending the LE Set Host Feature command. The command is sent to indicate host support for Connected Isochronous Streams in this case. It has been observed that the system could not initialize BIS-only capable controllers because the controllers do not support the command.
As per Core v6.2 | Vol 4, Part E, Table 3.1 the command shall be supported if CIS Central or CIS Peripheral is supported; otherwise, the command is optional.
--- net/bluetooth/hci_sync.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
net/bluetooth/hci_sync.c +1 -1
@@ -4564,7 +4564,7 @@ static int hci_le_set_host_feature_sync(struct hci_dev *hdev) {struct hci_cp_le_set_host_feature cp; -if (!iso_capable (hdev)) +if (!cis_capable (hdev))return 0 ;memset (&cp,0 ,sizeof (cp));
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 174/311 ipmi: Fix use-after-free and list corruption on sender error
In reply to Sasha LevinFrom: Corey Minyard <[email protected]>
commit 594c11d0e1d445f580898a2b8c850f2e3f099368 upstream.
The analysis from Breno:
When the SMI sender returns an error, smi_work() delivers an error response but then jumps back to restart without cleaning up properly:
1. intf->curr_msg is not cleared, so no new message is pulled
2. newmsg still points to the message, causing sender() to be called again with the same message
3. If sender() fails again, deliver_err_response() is called with the same recv_msg that was already queued for delivery
This causes list_add corruption ("list_add double add") because the recv_msg is added to the user_msgs list twice. Subsequently, the corrupted list leads to use-after-free when the memory is freed and reused, and eventually a NULL pointer dereference when accessing recv_msg->done.
The buggy sequence:
sender() fails
-> deliver_err_response(recv_msg) // recv_msg queued for delivery
-> goto restart // curr_msg not cleared!
sender() fails again (same message!)
-> deliver_err_response(recv_msg) // tries to queue same recv_msg
-> LIST CORRUPTIONFix this by freeing the message and setting it to NULL on a send error. Also, always free the newmsg on a send error, otherwise it will leak.
--- drivers/char/ipmi/ipmi_msghandler.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-)
drivers/char/ipmi/ipmi_msghandler.c +9 -2
@@ -4852,8 +4852,15 @@ static void smi_work(struct work_struct *t)if (newmsg->recv_msg)deliver_err_response (intf, newmsg->recv_msg, cc); -else -ipmi_free_smi_msg (newmsg); +if (!run_to_completion) +spin_lock_irqsave (&intf->xmit_msgs_lock, + flags); + intf->curr_msg =NULL ; +if (!run_to_completion) +spin_unlock_irqrestore (&intf->xmit_msgs_lock, + flags); +ipmi_free_smi_msg (newmsg); + newmsg =NULL ; goto restart; } }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 175/311 net: stmmac: remove support for lpi_intr_o
In reply to Sasha LevinFrom: "Russell King (Oracle)" <[email protected]>
commit 14eb64db8ff07b58a35b98375f446d9e20765674 upstream.
The dwmac databook for v3.74a states that lpi_intr_o is a sideband signal which should be used to ungate the application clock, and this signal is synchronous to the receive clock. The receive clock can run at 2.5, 25 or 125MHz depending on the media speed, and can stop under the control of the link partner. This means that the time it takes to clear is dependent on the negotiated media speed, and thus can be 8, 40, or 400ns after reading the LPI control and status register.
It has been observed with some aggressive link partners, this clock can stop while lpi_intr_o is still asserted, meaning that the signal remains asserted for an indefinite period that the local system has no direct control over.
The LPI interrupts will still be signalled through the main interrupt path in any case, and this path is not dependent on the receive clock.
This, since we do not gate the application clock, and the chances of adding clock gating in the future are slim due to the clocks being ill-defined, lpi_intr_o serves no useful purpose. Remove the code which requests the interrupt, and all associated code.
--- drivers/net/ethernet/stmicro/stmmac/common.h | 1 - .../net/ethernet/stmicro/stmmac/dwmac-intel.c | 4 --- .../ethernet/stmicro/stmmac/dwmac-loongson.c | 7 ---- drivers/net/ethernet/stmicro/stmmac/stmmac.h | 2 -- .../net/ethernet/stmicro/stmmac/stmmac_main.c | 36 ------------------- .../ethernet/stmicro/stmmac/stmmac_platform.c | 8 ----- include/linux/stmmac.h | 1 - 7 files changed, 59 deletions(-)
drivers/net/ethernet/stmicro/stmmac/common.h +0 -1
@@ -390,7 +390,6 @@ enum request_irq_err {REQ_IRQ_ERR_SFTY ,REQ_IRQ_ERR_SFTY_UE ,REQ_IRQ_ERR_SFTY_CE , -REQ_IRQ_ERR_LPI ,REQ_IRQ_ERR_WOL ,REQ_IRQ_ERR_MAC , REQ_IRQ_ERR_NO,
drivers/net/ethernet/stmicro/stmmac/dwmac-intel.c +0 -4
@@ -719,7 +719,6 @@ static int intel_mgbe_common_data(struct pci_dev *pdev,/* Setup MSI vector offset specific to Intel mGbE controller */ plat->msi_mac_vec =29 ; - plat->msi_lpi_vec =28 ; plat->msi_sfty_ce_vec =27 ; plat->msi_sfty_ue_vec =26 ; plat->msi_rx_base_vec =0 ;
@@ -1177,8 +1176,6 @@ static int stmmac_config_multi_msi(struct pci_dev *pdev, res->irq =pci_irq_vector (pdev, plat->msi_mac_vec);if (plat->msi_wol_vec <STMMAC_MSI_VEC_MAX ) res->wol_irq =pci_irq_vector (pdev, plat->msi_wol_vec); -if (plat->msi_lpi_vec <STMMAC_MSI_VEC_MAX ) - res->lpi_irq =pci_irq_vector (pdev, plat->msi_lpi_vec);if (plat->msi_sfty_ce_vec <STMMAC_MSI_VEC_MAX ) res->sfty_ce_irq =pci_irq_vector (pdev, plat->msi_sfty_ce_vec);if (plat->msi_sfty_ue_vec <STMMAC_MSI_VEC_MAX )
@@ -1294,7 +1291,6 @@ static int intel_eth_pci_probe(struct pci_dev *pdev,*/ plat->msi_mac_vec =STMMAC_MSI_VEC_MAX ; plat->msi_wol_vec =STMMAC_MSI_VEC_MAX ; - plat->msi_lpi_vec =STMMAC_MSI_VEC_MAX ; plat->msi_sfty_ce_vec =STMMAC_MSI_VEC_MAX ; plat->msi_sfty_ue_vec =STMMAC_MSI_VEC_MAX ; plat->msi_rx_base_vec =STMMAC_MSI_VEC_MAX ;
drivers/net/ethernet/stmicro/stmmac/dwmac-loongson.c +0 -7
@@ -443,13 +443,6 @@ static int loongson_dwmac_dt_config(struct pci_dev *pdev, res->wol_irq = res->irq; } - res->lpi_irq =of_irq_get_byname (np,"eth_lpi" ); -if (res->lpi_irq <0 ) { -dev_err (&pdev->dev,"IRQ eth_lpi not found\n" ); - ret = -ENODEV ; - goto err_put_node; - } - ret =device_get_phy_mode (&pdev->dev);if (ret <0 ) {dev_err (&pdev->dev,"phy_mode not found\n" );
drivers/net/ethernet/stmicro/stmmac/stmmac.h +0 -2
@@ -31,7 +31,6 @@ struct stmmac_resources {void __iomem *addr;u8 mac[ETH_ALEN ];int wol_irq; -int lpi_irq;int irq;int sfty_irq;int sfty_ce_irq;
@@ -297,7 +296,6 @@ struct stmmac_priv {int wol_irq;u32 gmii_address_bus_config;struct timer_list eee_ctrl_timer; -int lpi_irq;u32 tx_lpi_timer;bool tx_lpi_clk_stop;bool eee_enabled;
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +0 -36
@@ -3712,10 +3712,6 @@ static void stmmac_free_irq(struct net_device *dev,free_irq (priv->sfty_ce_irq, dev); fallthrough;case REQ_IRQ_ERR_SFTY_CE : -if (priv->lpi_irq >0 && priv->lpi_irq != dev->irq) -free_irq (priv->lpi_irq, dev); - fallthrough; -case REQ_IRQ_ERR_LPI :if (priv->wol_irq >0 && priv->wol_irq != dev->irq)free_irq (priv->wol_irq, dev); fallthrough;
@@ -3773,24 +3769,6 @@ static int stmmac_request_irq_multi_msi(struct net_device *dev) } } -/* Request the LPI IRQ in case of another line -* is used for LPI -*/ -if (priv->lpi_irq >0 && priv->lpi_irq != dev->irq) { - int_name = priv->int_name_lpi; -sprintf (int_name,"%s:%s" , dev->name,"lpi" ); - ret =request_irq (priv->lpi_irq, - stmmac_mac_interrupt, -0 , int_name, dev); -if (unlikely (ret <0 )) { -netdev_err (priv->dev, -"%s: alloc lpi MSI %d (error: %d)\n" , - __func__, priv->lpi_irq, ret); - irq_err =REQ_IRQ_ERR_LPI ; - goto irq_error; - } - } -/* Request the common Safety Feature Correctible/Uncorrectible * Error line in case of another line is used */
@@ -3930,19 +3908,6 @@ static int stmmac_request_irq_single(struct net_device *dev) } } -/* Request the IRQ lines */ -if (priv->lpi_irq >0 && priv->lpi_irq != dev->irq) { - ret =request_irq (priv->lpi_irq, stmmac_interrupt, -IRQF_SHARED , dev->name, dev); -if (unlikely (ret <0 )) { -netdev_err (priv->dev, -"%s: ERROR: allocating the LPI IRQ %d (%d)\n" , - __func__, priv->lpi_irq, ret); - irq_err =REQ_IRQ_ERR_LPI ; - goto irq_error; - } - } -/* Request the common Safety Feature Correctible/Uncorrectible * Error line in case of another line is used */
@@ -7709,7 +7674,6 @@ static int __stmmac_dvr_probe(struct device *device, priv->dev->irq = res->irq; priv->wol_irq = res->wol_irq; - priv->lpi_irq = res->lpi_irq; priv->sfty_irq = res->sfty_irq; priv->sfty_ce_irq = res->sfty_ce_irq; priv->sfty_ue_irq = res->sfty_ue_irq;
drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c +0 -8
@@ -725,14 +725,6 @@ int stmmac_get_platform_resources(struct platform_device *pdev, stmmac_res->wol_irq = stmmac_res->irq; } - stmmac_res->lpi_irq = -platform_get_irq_byname_optional (pdev,"eth_lpi" ); -if (stmmac_res->lpi_irq <0 ) { -if (stmmac_res->lpi_irq == -EPROBE_DEFER ) -return -EPROBE_DEFER ; -dev_info (&pdev->dev,"IRQ eth_lpi not found\n" ); - } - stmmac_res->sfty_irq =platform_get_irq_byname_optional (pdev,"sfty" );if (stmmac_res->sfty_irq <0 ) {
include/linux/stmmac.h +0 -1
@@ -299,7 +299,6 @@ struct plat_stmmacenet_data {int int_snapshot_num;int msi_mac_vec;int msi_wol_vec; -int msi_lpi_vec;int msi_sfty_ce_vec;int msi_sfty_ue_vec;int msi_rx_base_vec;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 176/311 drm/amd/display: Use GFP_ATOMIC in dc_create_stream_for_sink
In reply to Sasha LevinFrom: Natalie Vock <[email protected]>
commit 28dfe4317541e57fe52f9a290394cd29c348228b upstream.
This can be called while preemption is disabled, for example by dcn32_internal_validate_bw which is called with the FPU active.
Fixes "BUG: scheduling while atomic" messages I encounter on my Navi31 machine.
--- drivers/gpu/drm/amd/display/dc/core/dc_stream.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/amd/display/dc/core/dc_stream.c +1 -1
@@ -169,7 +169,7 @@ struct dc_stream_state *dc_create_stream_for_sink(if (sink ==NULL )return NULL ; - stream =kzalloc (sizeof (struct dc_stream_state ),GFP_KERNEL ); + stream =kzalloc (sizeof (struct dc_stream_state ),GFP_ATOMIC );if (stream ==NULL ) goto alloc_fail;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 177/311 nvme: fix admin queue leak on controller reset
In reply to Sasha LevinFrom: Ming Lei <[email protected]>
[ Upstream commit b84bb7bd913d8ca2f976ee6faf4a174f91c02b8d ]
When nvme_alloc_admin_tag_set() is called during a controller reset, a previous admin queue may still exist. Release it properly before allocating a new one to avoid orphaning the old queue.
This fixes a regression introduced by commit 03b3bcd319b3 ("nvme: fix admin request_queue lifetime").
--- drivers/nvme/host/core.c | 7 +++++++ 1 file changed, 7 insertions(+)
drivers/nvme/host/core.c +7 -0
@@ -4864,6 +4864,13 @@ int nvme_alloc_admin_tag_set(struct nvme_ctrl *ctrl, struct blk_mq_tag_set *set,if (ret)return ret; +/* +* If a previous admin queue exists (e.g., from before a reset), +* put it now before allocating a new one to avoid orphaning it. +*/ +if (ctrl->admin_q) +blk_put_queue (ctrl->admin_q); + ctrl->admin_q =blk_mq_alloc_queue (set, &lim,NULL );if (IS_ERR (ctrl->admin_q)) { ret =PTR_ERR (ctrl->admin_q);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 178/311 hwmon: (macsmc) Fix regressions in Apple Silicon SMC hwmon driver
In reply to Sasha LevinFrom: Guenter Roeck <[email protected]>
[ Upstream commit 5dd69b864911ae3847365e8bafe7854e79fbeecb ]
The recently added macsmc-hwmon driver contained several critical bugs in its sensor population logic and float conversion routines.
Specifically:
- The voltage sensor population loop used the wrong prefix ("volt-" instead of "voltage-") and incorrectly assigned sensors to the temperature sensor array (hwmon->temp.sensors) instead of the voltage sensor array (hwmon->volt.sensors). This would lead to out-of-bounds memory access or data corruption when both temperature and voltage sensors were present.
- The float conversion in macsmc_hwmon_write_f32() had flawed exponent logic for values >= 2^24 and lacked masking for the mantissa, which could lead to incorrect values being written to the SMC.
Fix these issues to ensure correct sensor registration and reliable manual fan control.
Confirm that the reported overflow in FIELD_PREP is fixed by declaring macsmc_hwmon_write_f32() as __always_inline for a compile test.
--- drivers/hwmon/macsmc-hwmon.c | 25 +++++++++++-------------- 1 file changed, 11 insertions(+), 14 deletions(-)
drivers/hwmon/macsmc-hwmon.c +11 -14
@@ -228,25 +228,22 @@ static int macsmc_hwmon_write_f32(struct apple_smc *smc, smc_key key, int value) {u64 val;u32 fval =0 ; -int exp =0 , neg; +int exp, neg; + neg = value <0 ; val =abs (value); - neg = val != value;if (val) { -int msb =__fls (val) - exp; - -if (msb >23 ) { - val >>= msb -FLT_MANT_BIAS ; - exp -= msb -FLT_MANT_BIAS ; - }else if (msb <23 ) { - val <<=FLT_MANT_BIAS - msb; - exp += msb; - } + exp =__fls (val); + +if (exp >23 ) + val >>= exp -23 ; +else + val <<=23 - exp; fval =FIELD_PREP (FLT_SIGN_MASK , neg) |FIELD_PREP (FLT_EXP_MASK , exp +FLT_EXP_BIAS ) | -FIELD_PREP (FLT_MANT_MASK , val); +FIELD_PREP (FLT_MANT_MASK , val &FLT_MANT_MASK ); }return apple_smc_write_u32 (smc, key, fval);
@@ -663,8 +660,8 @@ static int macsmc_hwmon_populate_sensors(struct macsmc_hwmon *hwmon,if (!hwmon->volt.sensors)return -ENOMEM ; -for_each_child_of_node_with_prefix (hwmon_node, key_node,"volt-" ) { - sensor = &hwmon->temp.sensors[hwmon->temp.count]; +for_each_child_of_node_with_prefix (hwmon_node, key_node,"voltage-" ) { + sensor = &hwmon->volt.sensors[hwmon->volt.count];if (!macsmc_hwmon_create_sensor (hwmon->dev, hwmon->smc, key_node, sensor)) { sensor->attrs =HWMON_I_INPUT ;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 179/311 hwmon: (macsmc) Fix overflows, underflows, and sign extension
In reply to Sasha LevinFrom: Guenter Roeck <[email protected]>
[ Upstream commit 579b86f3c26fee97996e68c1cbfb7461711f3de3 ]
The macsmc-hwmon driver experienced several issues related to value scaling and type conversion:
1. macsmc_hwmon_read_f32_scaled() clipped values to INT_MAX/INT_MIN. On 64-bit systems, hwmon supports long values, so clipping to 32-bit range was premature and caused loss of range for high-power sensors. Changed it to use long and clip to LONG_MAX/LONG_MIN.
2. The overflow check in macsmc_hwmon_read_f32_scaled() used 1UL, which is 32-bit on some platforms. Switched to 1ULL.
3. macsmc_hwmon_read_key() used a u32 temporary variable for f32 values. When assigned to a 64-bit long, negative values were zero-extended instead of sign-extended, resulting in large positive numbers.
4. macsmc_hwmon_read_ioft_scaled() used mult_frac() which could overflow during intermediate multiplication. Switched to mul_u64_u32_div() to handle the 64-bit multiplication safely.
5. ioft values (unsigned 48.16) could overflow long when scaled by 1,000,000. Added explicit clipping to LONG_MAX in the caller.
6. macsmc_hwmon_write_f32() truncated its long argument to int, potentially causing issues for large values.
Fix these issues by using appropriate types and helper functions.
--- drivers/hwmon/macsmc-hwmon.c | 28 ++++++++++++++++------------ 1 file changed, 16 insertions(+), 12 deletions(-)
drivers/hwmon/macsmc-hwmon.c +16 -12
@@ -22,6 +22,7 @@#include <linux/bitfield.h> #include <linux/hwmon.h> +#include <linux/math64.h> #include <linux/mfd/macsmc.h> #include <linux/module.h> #include <linux/of.h>
@@ -130,7 +131,7 @@ static int macsmc_hwmon_read_ioft_scaled(struct apple_smc *smc, smc_key key,if (ret <0 )return ret; - *p =mult_frac (val, scale,65536 ); + *p =mul_u64_u32_div (val, scale,65536 );return 0 ; }
@@ -140,7 +141,7 @@ static int macsmc_hwmon_read_ioft_scaled(struct apple_smc *smc, smc_key key,* them. */ static int macsmc_hwmon_read_f32_scaled (struct apple_smc *smc,smc_key key, -int *p,int scale) +long *p,int scale) {u32 fval;u64 val;
@@ -162,21 +163,21 @@ static int macsmc_hwmon_read_f32_scaled(struct apple_smc *smc, smc_key key, val =0 ; else if (exp < 0) val >>= -exp; - else if (exp != 0 && (val & ~((1UL << (64 - exp)) - 1))) /* overflow */ + else if (exp != 0 && (val & ~((1ULL << (64 - exp)) - 1))) /* overflow */ val =U64_MAX ; else val <<= exp; if (fval & FLT_SIGN_MASK) { -if (val > (-(s64 )INT_MIN )) - *p =INT_MIN ; +if (val > (u64 )LONG_MAX +1 ) + *p =LONG_MIN ;else - *p = -val; + *p = -(long )val; } else { -if (val >INT_MAX ) - *p =INT_MAX ; +if (val > (u64 )LONG_MAX ) + *p =LONG_MAX ;else - *p = val; + *p = (long )val; }return 0 ;
@@ -195,7 +196,7 @@ static int macsmc_hwmon_read_key(struct apple_smc *smc,switch (sensor->info.type_code) {/* 32-bit IEEE 754 float */ case __SMC_KEY ('f' ,'l' ,'t' ,' ' ): { -u32 flt_ =0 ; +long flt_ =0 ; ret =macsmc_hwmon_read_f32_scaled (smc, sensor->macsmc_key, &flt_, scale);
@@ -214,7 +215,10 @@ static int macsmc_hwmon_read_key(struct apple_smc *smc,if (ret)return ret; - *val = (long )ioft; +if (ioft >LONG_MAX ) + *val =LONG_MAX ; +else + *val = (long )ioft;break ; }default :
@@ -224,7 +228,7 @@ static int macsmc_hwmon_read_key(struct apple_smc *smc,return 0 ; } -static int macsmc_hwmon_write_f32 (struct apple_smc *smc,smc_key key,int value) +static int macsmc_hwmon_write_f32 (struct apple_smc *smc,smc_key key,long value) {u64 val;u32 fval =0 ;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 180/311 hwmon: (aht10) Fix initialization commands for AHT20
In reply to Sasha LevinFrom: Hao Yu <[email protected]>
[ Upstream commit b7497b5a99f54ab8dcda5b14a308385b2fb03d8d ]
According to the AHT20 datasheet (updated to V1.0 after the 2023.09 version), the initialization command for AHT20 is 0b10111110 (0xBE). The previous sequence (0xE1) used in earlier versions is no longer compatible with newer AHT20 sensors. Update the initialization command to ensure the sensor is properly initialized.
While at it, use binary notation for DHT20_CMD_INIT to match the notation used in the datasheet.
--- drivers/hwmon/aht10.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/hwmon/aht10.c +4 -2
@@ -37,7 +37,9 @@#define AHT10_CMD_MEAS 0b10101100#define AHT10_CMD_RST 0b10111010 -#define DHT20_CMD_INIT 0x71 +#define AHT20_CMD_INIT 0b10111110 + +#define DHT20_CMD_INIT 0b01110001/* * Flags in the answer byte/command
@@ -341,7 +343,7 @@ static int aht10_probe(struct i2c_client *client) data->meas_size =AHT20_MEAS_SIZE ; data->crc8 = true;crc8_populate_msb (crc8_table,AHT20_CRC8_POLY ); - data->init_cmd =AHT10_CMD_INIT ; + data->init_cmd =AHT20_CMD_INIT ;break ;case dht20: data->meas_size =AHT20_MEAS_SIZE ;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 181/311 pinctrl: equilibrium: rename irq_chip function callbacks
In reply to Sasha LevinFrom: Florian Eckert <[email protected]>
[ Upstream commit 1f96b84835eafb3e6f366dc3a66c0e69504cec9d ]
Renaming of the irq_chip callback functions to improve clarity.
Stable-dep-of: 3e00b1b332e5 ("pinctrl: equilibrium: fix warning trace on load")
--- drivers/pinctrl/pinctrl-equilibrium.c | 24 ++++++++++++------------ 1 file changed, 12 insertions(+), 12 deletions(-)
drivers/pinctrl/pinctrl-equilibrium.c +12 -12
@@ -23,7 +23,7 @@#define PIN_NAME_LEN 10#define PAD_REG_OFF 0x100 -static void eqbr_gpio_disable_irq (struct irq_data *d) +static void eqbr_irq_mask (struct irq_data *d) {struct gpio_chip *gc =irq_data_get_irq_chip_data (d);struct eqbr_gpio_ctrl *gctrl =gpiochip_get_data (gc);
@@ -36,7 +36,7 @@ static void eqbr_gpio_disable_irq(struct irq_data *d)gpiochip_disable_irq (gc, offset); } -static void eqbr_gpio_enable_irq (struct irq_data *d) +static void eqbr_irq_unmask (struct irq_data *d) {struct gpio_chip *gc =irq_data_get_irq_chip_data (d);struct eqbr_gpio_ctrl *gctrl =gpiochip_get_data (gc);
@@ -50,7 +50,7 @@ static void eqbr_gpio_enable_irq(struct irq_data *d)raw_spin_unlock_irqrestore (&gctrl->lock, flags); } -static void eqbr_gpio_ack_irq (struct irq_data *d) +static void eqbr_irq_ack (struct irq_data *d) {struct gpio_chip *gc =irq_data_get_irq_chip_data (d);struct eqbr_gpio_ctrl *gctrl =gpiochip_get_data (gc);
@@ -62,10 +62,10 @@ static void eqbr_gpio_ack_irq(struct irq_data *d)raw_spin_unlock_irqrestore (&gctrl->lock, flags); } -static void eqbr_gpio_mask_ack_irq (struct irq_data *d) +static void eqbr_irq_mask_ack (struct irq_data *d) { -eqbr_gpio_disable_irq (d); -eqbr_gpio_ack_irq (d); +eqbr_irq_mask (d); +eqbr_irq_ack (d); } static inline void eqbr_cfg_bit(void __iomem *addr,
@@ -92,7 +92,7 @@ static int eqbr_irq_type_cfg(struct gpio_irq_type *type,return 0 ; } -static int eqbr_gpio_set_irq_type (struct irq_data *d,unsigned int type) +static int eqbr_irq_set_type (struct irq_data *d,unsigned int type) {struct gpio_chip *gc =irq_data_get_irq_chip_data (d);struct eqbr_gpio_ctrl *gctrl =gpiochip_get_data (gc);
@@ -166,11 +166,11 @@ static void eqbr_irq_handler(struct irq_desc *desc)static const struct irq_chip eqbr_irq_chip = { .name ="gpio_irq" , - .irq_mask = eqbr_gpio_disable_irq, - .irq_unmask = eqbr_gpio_enable_irq, - .irq_ack = eqbr_gpio_ack_irq, - .irq_mask_ack = eqbr_gpio_mask_ack_irq, - .irq_set_type = eqbr_gpio_set_irq_type, + .irq_ack = eqbr_irq_ack, + .irq_mask = eqbr_irq_mask, + .irq_mask_ack = eqbr_irq_mask_ack, + .irq_unmask = eqbr_irq_unmask, + .irq_set_type = eqbr_irq_set_type, .flags =IRQCHIP_IMMUTABLE ,GPIOCHIP_IRQ_RESOURCE_HELPERS , };
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 182/311 pinctrl: equilibrium: fix warning trace on load
In reply to Sasha LevinFrom: Florian Eckert <[email protected]>
[ Upstream commit 3e00b1b332e54ba50cca6691f628b9c06574024f ]
The callback functions 'eqbr_irq_mask()' and 'eqbr_irq_ack()' are also called in the callback function 'eqbr_irq_mask_ack()'. This is done to avoid source code duplication. The problem, is that in the function 'eqbr_irq_mask()' also calles the gpiolib function 'gpiochip_disable_irq()'
This generates the following warning trace in the log for every gpio on load.
[ 6.088111] ------------[ cut here ]------------ [ 6.092440] WARNING: CPU: 3 PID: 1 at drivers/gpio/gpiolib.c:3810 gpiochip_disable_irq+0x39/0x50 [ 6.097847] Modules linked in: [ 6.097847] CPU: 3 UID: 0 PID: 1 Comm: swapper/0 Tainted: G W 6.12.59+ #0 [ 6.097847] Tainted: [W]=WARN [ 6.097847] RIP: 0010:gpiochip_disable_irq+0x39/0x50 [ 6.097847] Code: 39 c6 48 19 c0 21 c6 48 c1 e6 05 48 03 b2 38 03 00 00 48 81 fe 00 f0 ff ff 77 11 48 8b 46 08 f6 c4 02 74 06 f0 80 66 09 fb c3 <0f> 0b 90 0f 1f 40 00 c3 66 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 [ 6.097847] RSP: 0000:ffffc9000000b830 EFLAGS: 00010046 [ 6.097847] RAX: 0000000000000045 RBX: ffff888001be02a0 RCX: 0000000000000008 [ 6.097847] RDX: ffff888001be9000 RSI: ffff888001b2dd00 RDI: ffff888001be02a0 [ 6.097847] RBP: ffffc9000000b860 R08: 0000000000000000 R09: 0000000000000000 [ 6.097847] R10: 0000000000000001 R11: ffff888001b2a154 R12: ffff888001be0514 [ 6.097847] R13: ffff888001be02a0 R14: 0000000000000008 R15: 0000000000000000 [ 6.097847] FS: 0000000000000000(0000) GS:ffff888041d80000(0000) knlGS:0000000000000000 [ 6.097847] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 6.097847] CR2: 0000000000000000 CR3: 0000000003030000 CR4: 00000000001026b0 [ 6.097847] Call Trace: [ 6.097847] <TASK> [ 6.097847] ? eqbr_irq_mask+0x63/0x70 [ 6.097847] ? no_action+0x10/0x10 [ 6.097847] eqbr_irq_mask_ack+0x11/0x60
In an other driver (drivers/pinctrl/starfive/pinctrl-starfive-jh7100.c) the interrupt is not disabled here.
To fix this, do not call the 'eqbr_irq_mask()' and 'eqbr_irq_ack()' function. Implement instead this directly without disabling the interrupts.
--- drivers/pinctrl/pinctrl-equilibrium.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-)
drivers/pinctrl/pinctrl-equilibrium.c +9 -2
@@ -64,8 +64,15 @@ static void eqbr_irq_ack(struct irq_data *d)static void eqbr_irq_mask_ack (struct irq_data *d) { -eqbr_irq_mask (d); -eqbr_irq_ack (d); +struct gpio_chip *gc =irq_data_get_irq_chip_data (d); +struct eqbr_gpio_ctrl *gctrl =gpiochip_get_data (gc); +unsigned int offset =irqd_to_hwirq (d); +unsigned long flags; + +raw_spin_lock_irqsave (&gctrl->lock, flags); +writel (BIT (offset), gctrl->membase +GPIO_IRNENCLR ); +writel (BIT (offset), gctrl->membase +GPIO_IRNCR ); +raw_spin_unlock_irqrestore (&gctrl->lock, flags); } static inline void eqbr_cfg_bit(void __iomem *addr,
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 183/311 pinctrl: qcom: qcs615: Add missing dual edge GPIO IRQ errata flag
In reply to Sasha LevinFrom: Maulik Shah <[email protected]>
[ Upstream commit 09a30b7a035f9f4ac918c8a9af89d70e43462152 ]
Wakeup capable GPIOs uses PDC as parent IRQ chip and PDC on qcs615 do not support dual edge IRQs. Add missing wakeirq_dual_edge_errata configuration to enable workaround for dual edge GPIO IRQs.
--- drivers/pinctrl/qcom/pinctrl-qcs615.c | 1 + 1 file changed, 1 insertion(+)
drivers/pinctrl/qcom/pinctrl-qcs615.c +1 -0
@@ -1067,6 +1067,7 @@ static const struct msm_pinctrl_soc_data qcs615_tlmm = { .ntiles = ARRAY_SIZE(qcs615_tiles), .wakeirq_map = qcs615_pdc_map, .nwakeirq_map = ARRAY_SIZE(qcs615_pdc_map), + .wakeirq_dual_edge_errata = true, };static const struct of_device_id qcs615_tlmm_of_match[] = {
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 184/311 platform/x86: thinkpad_acpi: Fix errors reading battery thresholds
In reply to Sasha LevinFrom: Jonathan Teh <[email protected]>
[ Upstream commit 53e977b1d50c46f2c4ec3865cd13a822f58ad3cd ]
Check whether the battery supports the relevant charge threshold before reading the value to silence these errors:
thinkpad_acpi: acpi_evalf(BCTG, dd, ...) failed: AE_NOT_FOUND
ACPI: \_SB_.PCI0.LPC_.EC__.HKEY: BCTG: evaluate failedthinkpad_acpi: acpi_evalf(BCSG, dd, ...) failed: AE_NOT_FOUND
ACPI: \_SB_.PCI0.LPC_.EC__.HKEY: BCSG: evaluate failed
when reading the charge thresholds via sysfs on platforms that do not support them such as the ThinkPad T400.
--- drivers/platform/x86/lenovo/thinkpad_acpi.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/platform/x86/lenovo/thinkpad_acpi.c +4 -2
@@ -9525,14 +9525,16 @@ static int tpacpi_battery_get(int what, int battery, int *ret) {switch (what) {case THRESHOLD_START : - if ACPI_FAILURE(tpacpi_battery_acpi_eval(GET_START, ret, battery)) +if (!battery_info.batteries[battery].start_support || +ACPI_FAILURE (tpacpi_battery_acpi_eval (GET_START , ret, battery)))return -ENODEV ;/* The value is in the low 8 bits of the response */ *ret = *ret &0xFF ;return 0 ;case THRESHOLD_STOP : - if ACPI_FAILURE(tpacpi_battery_acpi_eval(GET_STOP, ret, battery)) +if (!battery_info.batteries[battery].stop_support || +ACPI_FAILURE (tpacpi_battery_acpi_eval (GET_STOP , ret, battery)))return -ENODEV ;/* Value is in lower 8 bits */ *ret = *ret &0xFF ;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 185/311 module: Remove duplicate freeing of lockdep classes
In reply to Sasha LevinFrom: Petr Pavlu <[email protected]>
[ Upstream commit a7b4bc094fbaa7dc7b7b91ae33549bbd7eefaac1 ]
In the error path of load_module(), under the free_module label, the code calls lockdep_free_key_range() to release lock classes associated with the MOD_DATA, MOD_RODATA and MOD_RO_AFTER_INIT module regions, and subsequently invokes module_deallocate().
Since commit ac3b43283923 ("module: replace module_layout with module_memory"), the module_deallocate() function calls free_mod_mem(), which releases the lock classes as well and considers all module regions.
Attempting to free these classes twice is unnecessary. Remove the redundant code in load_module().
--- kernel/module/main.c | 6 ------ 1 file changed, 6 deletions(-)
kernel/module/main.c +0 -6
@@ -3544,12 +3544,6 @@ static int load_module(struct load_info *info, const char __user *uargs,mutex_unlock (&module_mutex); free_module:mod_stat_bump_invalid (info, flags); -/* Free lock-classes; relies on the preceding sync_rcu() */ - for_class_mod_mem_type(type, core_data) { -lockdep_free_key_range (mod->mem[type].base, - mod->mem[type].size); - } -module_memory_restore_rox (mod);module_deallocate (mod, info); free_copy:
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 186/311 HID: multitouch: new class MT_CLS_EGALAX_P80H84
In reply to Sasha LevinFrom: Ian Ray <[email protected]>
[ Upstream commit a2e70a89fa58133521b2deae4427d35776bda935 ]
--- drivers/hid/hid-multitouch.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-)
drivers/hid/hid-multitouch.c +9 -2
@@ -235,6 +235,7 @@ static void mt_post_parse(struct mt_device *td, struct mt_application *app);#define MT_CLS_SMART_TECH 0x0113#define MT_CLS_APPLE_TOUCHBAR 0x0114#define MT_CLS_YOGABOOK9I 0x0115 +#define MT_CLS_EGALAX_P80H84 0x0116#define MT_CLS_SIS 0x0457#define MT_DEFAULT_MAXCONTACT 10
@@ -449,6 +450,11 @@ static const struct mt_class mt_classes[] = { MT_QUIRK_YOGABOOK9I, .export_all_inputs = true }, + { .name = MT_CLS_EGALAX_P80H84, + .quirks =MT_QUIRK_ALWAYS_VALID | +MT_QUIRK_IGNORE_DUPLICATES | + MT_QUIRK_CONTACT_CNT_ACCURATE, + }, { } };
@@ -2233,8 +2239,9 @@ static const struct hid_device_id mt_devices[] = { { .driver_data = MT_CLS_EGALAX_SERIAL,MT_USB_DEVICE (USB_VENDOR_ID_DWAV , USB_DEVICE_ID_DWAV_EGALAX_MULTITOUCH_C000) }, - { .driver_data = MT_CLS_EGALAX, -MT_USB_DEVICE (USB_VENDOR_ID_DWAV , + { .driver_data = MT_CLS_EGALAX_P80H84, +HID_DEVICE (HID_BUS_ANY ,HID_GROUP_MULTITOUCH_WIN_8 , +USB_VENDOR_ID_DWAV , USB_DEVICE_ID_DWAV_EGALAX_MULTITOUCH_C002) },/* Elan devices */
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 187/311 pinctrl: pinconf-generic: Fix memory leak in pinconf_generic_parse_dt_config()
In reply to Sasha LevinFrom: Felix Gu <[email protected]>
[ Upstream commit 7a648d598cb8e8c62af3f0e020a25820a3f3a9a7 ]
In pinconf_generic_parse_dt_config(), if parse_dt_cfg() fails, it returns directly. This bypasses the cleanup logic and results in a memory leak of the cfg buffer.
Fix this by jumping to the out label on failure, ensuring kfree(cfg) is called before returning.
--- drivers/pinctrl/pinconf-generic.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/pinctrl/pinconf-generic.c +2 -2
@@ -351,13 +351,13 @@ int pinconf_generic_parse_dt_config(struct device_node *np, ret =parse_dt_cfg (np, dt_params,ARRAY_SIZE (dt_params), cfg, &ncfg);if (ret) -return ret; + goto out;if (pctldev && pctldev->desc->num_custom_params && pctldev->desc->custom_params) { ret =parse_dt_cfg (np, pctldev->desc->custom_params, pctldev->desc->num_custom_params, cfg, &ncfg);if (ret) -return ret; + goto out; }/* no configs found at all */
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 188/311 pinctrl: generic: move function to amlogic-am4 driver
In reply to Sasha LevinFrom: Conor Dooley <[email protected]>
[ Upstream commit 9c5a40f2922a5a6d6b42e7b3d4c8e253918c07a1 ]
pinconf_generic_dt_node_to_map_pinmux() is not actually a generic function, and really belongs in the amlogic-am4 driver. There are three reasons why.
First, and least, of the reasons is that this function behaves differently to the other dt_node_to_map functions in a way that is not obvious from a first glance. This difference stems for the devicetree properties that the function is intended for use with, and how they are typically used. The other generic dt_node_to_map functions support platforms where the pins, groups and functions are described statically in the driver and require a function that will produce a mapping from dt nodes to these pre-established descriptions. No other code in the driver is require to be executed at runtime. pinconf_generic_dt_node_to_map_pinmux() on the other hand is intended for use with the pinmux property, where groups and functions are determined entirely from the devicetree. As a result, there are no statically defined groups and functions in the driver for this function to perform a mapping to. Other drivers that use the pinmux property (e.g. the k1) their dt_node_to_map function creates the groups and functions as the devicetree is parsed. Instead of that, pinconf_generic_dt_node_to_map_pinmux() requires that the devicetree is parsed twice, once by it and once at probe, so that the driver dynamically creates the groups and functions before the dt_node_to_map callback is executed. I don't believe this double parsing requirement is how developers would expect this to work and is not necessary given there are drivers that do not have this behaviour.
Secondly and thirdly, the function bakes in some assumptions that only really match the amlogic platform about how the devicetree is constructed. These, to me, are problematic for something that claims to be generic.
The other dt_node_to_map implementations accept a being called for either a node containing pin configuration properties or a node containing child nodes that each contain the configuration properties. IOW, they support the following two devicetree configurations:
| cfg {
| label: group {
| pinmux = <asjhdasjhlajskd>;
| config-item1;
| };
| };| label: cfg {
| group1 {
| pinmux = <dsjhlfka>;
| config-item2;
| };
| group2 {
| pinmux = <lsdjhaf>;
| config-item1;
| };
| };pinconf_generic_dt_node_to_map_pinmux() only supports the latter.
The other assumption about devicetree configuration that the function makes is that the labeled node's parent is a "function node". The amlogic driver uses these "function nodes" to create the functions at probe time, and pinconf_generic_dt_node_to_map_pinmux() finds the parent of the node it is operating on's name as part of the mapping. IOW, it requires that the devicetree look like:
| pinctrl@bla {
|
| func-foo {
| label: group-default {
| pinmuxes = <lskdf>;
| };
| };
| };and couldn't be used if the nodes containing the pinmux and configuration properties are children of the pinctrl node itself:
| pinctrl@bla {
|
| label: group-default {
| pinmuxes = <lskdf>;
| };
| };These final two reasons are mainly why I believe this is not suitable as a generic function, and should be moved into the driver that is the sole user and originator of the "generic" function.
Stable-dep-of: a2539b92e4b7 ("pinctrl: meson: amlogic-a4: Fix device node reference leak in aml_dt_node_to_map_pinmux()")
--- drivers/pinctrl/meson/pinctrl-amlogic-a4.c | 71 +++++++++++++++++++++- drivers/pinctrl/pinconf-generic.c | 69 --------------------- include/linux/pinctrl/pinconf-generic.h | 5 -- 3 files changed, 70 insertions(+), 75 deletions(-)
drivers/pinctrl/meson/pinctrl-amlogic-a4.c +70 -1
@@ -24,6 +24,7 @@#include <dt-bindings/pinctrl/amlogic,pinctrl.h> #include "../core.h" +#include "../pinctrl-utils.h" #include "../pinconf.h" #define gpio_chip_to_bank (chip) \
@@ -672,11 +673,79 @@ static void aml_pin_dbg_show(struct pinctrl_dev *pcdev, struct seq_file *s,seq_printf (s," %s" ,dev_name (pcdev->dev)); } +static int aml_dt_node_to_map_pinmux (struct pinctrl_dev *pctldev, +struct device_node *np, +struct pinctrl_map **map, +unsigned int *num_maps) +{ +struct device *dev = pctldev->dev; +struct device_node *pnode; +unsigned long *configs =NULL ; +unsigned int num_configs =0 ; +struct property *prop; +unsigned int reserved_maps; +int reserve; +int ret; + + prop =of_find_property (np,"pinmux" ,NULL ); +if (!prop) { +dev_info (dev,"Missing pinmux property\n" ); +return -ENOENT ; + } + + pnode =of_get_parent (np); +if (!pnode) { +dev_info (dev,"Missing function node\n" ); +return -EINVAL ; + } + + reserved_maps =0 ; + *map =NULL ; + *num_maps =0 ; + + ret =pinconf_generic_parse_dt_config (np, pctldev, &configs, + &num_configs); +if (ret <0 ) { +dev_err (dev,"%pOF: could not parse node property\n" , np); +return ret; + } + + reserve =1 ; +if (num_configs) + reserve++; + + ret =pinctrl_utils_reserve_map (pctldev, map, &reserved_maps, + num_maps, reserve); +if (ret <0 ) + goto exit; + + ret =pinctrl_utils_add_map_mux (pctldev, map, + &reserved_maps, num_maps, np->name, + pnode->name); +if (ret <0 ) + goto exit; + +if (num_configs) { + ret =pinctrl_utils_add_map_configs (pctldev, map, &reserved_maps, + num_maps, np->name, configs, + num_configs,PIN_MAP_TYPE_CONFIGS_GROUP ); +if (ret <0 ) + goto exit; + } + +exit: +kfree (configs); +if (ret) +pinctrl_utils_free_map (pctldev, *map, *num_maps); + +return ret; +} +static const struct pinctrl_ops aml_pctrl_ops = { .get_groups_count = aml_get_groups_count, .get_group_name = aml_get_group_name, .get_group_pins = aml_get_group_pins, - .dt_node_to_map = pinconf_generic_dt_node_to_map_pinmux, + .dt_node_to_map = aml_dt_node_to_map_pinmux, .dt_free_map = pinconf_generic_dt_free_map, .pin_dbg_show = aml_pin_dbg_show, };
drivers/pinctrl/pinconf-generic.c +0 -69
75 lines not on this page · raw message
include/linux/pinctrl/pinconf-generic.h +0 -5
@@ -250,9 +250,4 @@ static inline int pinconf_generic_dt_node_to_map_all(struct pinctrl_dev *pctldevreturn pinconf_generic_dt_node_to_map (pctldev, np_config, map, num_maps, PIN_MAP_TYPE_INVALID); } - -int pinconf_generic_dt_node_to_map_pinmux (struct pinctrl_dev *pctldev, -struct device_node *np, -struct pinctrl_map **map, -unsigned int *num_maps);#endif /* __LINUX_PINCTRL_PINCONF_GENERIC_H */
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 189/311 pinctrl: meson: amlogic-a4: Fix device node reference leak in aml_dt_node_to_map_pinmux()
In reply to Sasha LevinFrom: Felix Gu <[email protected]>
[ Upstream commit a2539b92e4b791c1ba482930b5e51b1591975461 ]
The of_get_parent() function returns a device_node with an incremented reference count.
Use the __free(device_node) cleanup attribute to ensure of_node_put() is automatically called when pnode goes out of scope, fixing a reference leak.
--- drivers/pinctrl/meson/pinctrl-amlogic-a4.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/pinctrl/meson/pinctrl-amlogic-a4.c +1 -2
@@ -679,7 +679,6 @@ static int aml_dt_node_to_map_pinmux(struct pinctrl_dev *pctldev, unsigned int *num_maps) {struct device *dev = pctldev->dev; -struct device_node *pnode;unsigned long *configs =NULL ;unsigned int num_configs =0 ;struct property *prop;
@@ -693,7 +692,7 @@ static int aml_dt_node_to_map_pinmux(struct pinctrl_dev *pctldev,return -ENOENT ; } - pnode =of_get_parent (np); + struct device_node *pnode __free(device_node) = of_get_parent(np);if (!pnode) {dev_info (dev,"Missing function node\n" );return -EINVAL ;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 190/311 pinctrl: cirrus: cs42l43: Fix double-put in cs42l43_pin_probe()
In reply to Sasha LevinFrom: Felix Gu <[email protected]>
[ Upstream commit fd5bed798f45eb3a178ad527b43ab92705faaf8a ]
devm_add_action_or_reset() already invokes the action on failure, so the explicit put causes a double-put.
--- drivers/pinctrl/cirrus/pinctrl-cs42l43.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/pinctrl/cirrus/pinctrl-cs42l43.c +2 -3
@@ -574,10 +574,9 @@ static int cs42l43_pin_probe(struct platform_device *pdev)if (child) { ret =devm_add_action_or_reset (&pdev->dev, cs42l43_fwnode_put, child); -if (ret) { -fwnode_handle_put (child); +if (ret)return ret; - } +if (!child->dev) child->dev = priv->dev; fwnode = child;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 191/311 hwmon: (it87) Check the it87_lock() return value
In reply to Sasha LevinFrom: Bart Van Assche <[email protected]>
[ Upstream commit 07ed4f05bbfd2bc014974dcc4297fd3aa1cb88c0 ]
Return early in it87_resume() if it87_lock() fails instead of ignoring the return value of that function. This patch suppresses a Clang thread-safety warning.
--- drivers/hwmon/it87.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)
drivers/hwmon/it87.c +4 -1
@@ -3590,10 +3590,13 @@ static int it87_resume(struct device *dev) {struct platform_device *pdev =to_platform_device (dev);struct it87_data *data =dev_get_drvdata (dev); +int err;it87_resume_sio (pdev); -it87_lock (data); + err =it87_lock (data); +if (err) +return err;it87_check_pwm (dev);it87_check_limit_regs (data);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 192/311 idpf: increment completion queue next_to_clean in sw marker wait routine
In reply to Sasha LevinFrom: Li Li <[email protected]>
[ Upstream commit 712896ac4bce38a965a1c175f6e7804ed0381334 ]
Currently, in idpf_wait_for_sw_marker_completion(), when an IDPF_TXD_COMPLT_SW_MARKER packet is found, the routine breaks out of the for loop and does not increment the next_to_clean counter. This causes the subsequent NAPI polls to run into the same IDPF_TXD_COMPLT_SW_MARKER packet again and print out the following:
[ 23.261341] idpf 0000:05:00.0 eth1: Unknown TX completion type: 5
Instead, we should increment next_to_clean regardless when an IDPF_TXD_COMPLT_SW_MARKER packet is found.
Tested: with the patch applied, we do not see the errors above from NAPI polls anymore.
--- drivers/net/ethernet/intel/idpf/idpf_txrx.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/net/ethernet/intel/idpf/idpf_txrx.c +3 -3
@@ -2326,7 +2326,7 @@ void idpf_wait_for_sw_marker_completion(const struct idpf_tx_queue *txq) do { struct idpf_splitq_4b_tx_compl_desc *tx_desc; - struct idpf_tx_queue *target; + struct idpf_tx_queue *target = NULL; u32 ctype_gen, id; tx_desc = flow ? &complq->comp[ntc].common :
@@ -2346,14 +2346,14 @@ void idpf_wait_for_sw_marker_completion(const struct idpf_tx_queue *txq) target = complq->txq_grp->txqs[id];idpf_queue_clear (SW_MARKER , target); -if (target == txq) -break ; next:if (unlikely (++ntc == complq->desc_count)) { ntc =0 ; gen_flag = !gen_flag; } +if (target == txq) +break ; }while (time_before (jiffies, timeout));idpf_queue_assign (GEN_CHK , complq, gen_flag);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 193/311 idpf: change IRQ naming to match netdev and ethtool queue numbering
In reply to Sasha LevinFrom: Brian Vazquez <[email protected]>
[ Upstream commit 1500a8662d2d41d6bb03e034de45ddfe6d7d362d ]
The code uses the vidx for the IRQ name but that doesn't match ethtool reporting nor netdev naming, this makes it hard to tune the device and associate queues with IRQs. Sequentially requesting irqs starting from '0' makes the output consistent.
This commit changes the interrupt numbering but preserves the name format, maintaining ABI compatibility. Existing tools relying on the old numbering are already non-functional, as they lack a useful correlation to the interrupts.
Before:
ethtool -L eth1 tx 1 combined 3
grep . /proc/irq/*/*idpf*/../smp_affinity_list /proc/irq/67/idpf-Mailbox-0/../smp_affinity_list:0-55,112-167 /proc/irq/68/idpf-eth1-TxRx-1/../smp_affinity_list:0 /proc/irq/70/idpf-eth1-TxRx-3/../smp_affinity_list:1 /proc/irq/71/idpf-eth1-TxRx-4/../smp_affinity_list:2 /proc/irq/72/idpf-eth1-Tx-5/../smp_affinity_list:3
ethtool -S eth1 | grep -v ': 0'
NIC statistics:
tx_q-0_pkts: 1002
tx_q-1_pkts: 2679
tx_q-2_pkts: 1113
tx_q-3_pkts: 1192 <----- tx_q-3 vs idpf-eth1-Tx-5
rx_q-0_pkts: 1143
rx_q-1_pkts: 3172
rx_q-2_pkts: 1074After:
ethtool -L eth1 tx 1 combined 3
grep . /proc/irq/*/*idpf*/../smp_affinity_list
/proc/irq/67/idpf-Mailbox-0/../smp_affinity_list:0-55,112-167 /proc/irq/68/idpf-eth1-TxRx-0/../smp_affinity_list:0 /proc/irq/70/idpf-eth1-TxRx-1/../smp_affinity_list:1 /proc/irq/71/idpf-eth1-TxRx-2/../smp_affinity_list:2 /proc/irq/72/idpf-eth1-Tx-3/../smp_affinity_list:3
ethtool -S eth1 | grep -v ': 0'
NIC statistics:
tx_q-0_pkts: 118
tx_q-1_pkts: 134
tx_q-2_pkts: 228
tx_q-3_pkts: 138 <--- tx_q-3 matches idpf-eth1-Tx-3
rx_q-0_pkts: 111
rx_q-1_pkts: 366
rx_q-2_pkts: 120--- drivers/net/ethernet/intel/idpf/idpf_txrx.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/intel/idpf/idpf_txrx.c +1 -1
@@ -4038,7 +4038,7 @@ static int idpf_vport_intr_req_irq(struct idpf_vport *vport)continue ; name =kasprintf (GFP_KERNEL ,"%s-%s-%s-%d" , drv_name, if_name, - vec_name, vidx); + vec_name, vector); err =request_irq (irq_num, idpf_vport_intr_clean_queues,0 , name, q_vector);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 194/311 idpf: Fix flow rule delete failure due to invalid validation
In reply to Sasha LevinFrom: Sreedevi Joshi <[email protected]>
[ Upstream commit 2c31557336a8e4d209ed8d4513cef2c0f15e7ef4 ]
When deleting a flow rule using "ethtool -N <dev> delete <location>", idpf_sideband_action_ena() incorrectly validates fsp->ring_cookie even though ethtool doesn't populate this field for delete operations. The uninitialized ring_cookie may randomly match RX_CLS_FLOW_DISC or RX_CLS_FLOW_WAKE, causing validation to fail and preventing legitimate rule deletions. Remove the unnecessary sideband action enable check and ring_cookie validation during delete operations since action validation is not required when removing existing rules.
--- drivers/net/ethernet/intel/idpf/idpf_ethtool.c | 3 --- 1 file changed, 3 deletions(-)
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +0 -3
@@ -307,9 +307,6 @@ static int idpf_del_flow_steer(struct net_device *netdev, vport_config = vport->adapter->vport_config[np->vport_idx]; user_config = &vport_config->user_config; -if (!idpf_sideband_action_ena (vport, fsp)) -return -EOPNOTSUPP ; - rule =kzalloc (struct_size (rule, rule_info,1 ),GFP_KERNEL );if (!rule)return -ENOMEM ;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 195/311 ice: recap the VSI and QoS info after rebuild
In reply to Sasha LevinFrom: Aaron Ma <[email protected]>
[ Upstream commit 6aa07e23dd3ccd35a0100c06fcb6b6c3b01e7965 ]
Fix IRDMA hardware initialization timeout (-110) after resume by separating VSI-dependent configuration from RDMA resource allocation, ensuring VSI is rebuilt before IRDMA accesses it.
After resume from suspend, IRDMA hardware initialization fails:
ice: IRDMA hardware initialization FAILED init_state=4 status=-110
Separate RDMA initialization into two phases:
1. ice_init_rdma() - Allocate resources only (no VSI/QoS access, no plug)
2. ice_rdma_finalize_setup() - Assign VSI/QoS info and plug device
This allows:
- ice_init_rdma() to stay in ice_resume() (mirrors ice_deinit_rdma() in ice_suspend())
- VSI assignment deferred until after ice_vsi_rebuild() completes
- QoS info updated after ice_dcb_rebuild() completes
- Device plugged only when control queues, VSI, and DCB are all ready
--- drivers/net/ethernet/intel/ice/ice.h | 1 + drivers/net/ethernet/intel/ice/ice_idc.c | 44 +++++++++++++++++------ drivers/net/ethernet/intel/ice/ice_main.c | 7 +++- 3 files changed, 41 insertions(+), 11 deletions(-)
drivers/net/ethernet/intel/ice/ice.h +1 -0
@@ -990,6 +990,7 @@ int ice_schedule_reset(struct ice_pf *pf, enum ice_reset_req reset);void ice_print_link_msg (struct ice_vsi *vsi,bool isup);int ice_plug_aux_dev (struct ice_pf *pf);void ice_unplug_aux_dev (struct ice_pf *pf); +void ice_rdma_finalize_setup (struct ice_pf *pf);int ice_init_rdma (struct ice_pf *pf);void ice_deinit_rdma (struct ice_pf *pf);bool ice_is_wol_supported (struct ice_hw *hw);
drivers/net/ethernet/intel/ice/ice_idc.c +34 -10
69 lines not on this page · raw message
drivers/net/ethernet/intel/ice/ice_main.c +6 -1
@@ -5138,6 +5138,9 @@ int ice_load(struct ice_pf *pf)if (err) goto err_init_rdma; +/* Finalize RDMA: VSI already created, assign info and plug device */ +ice_rdma_finalize_setup (pf); +ice_service_task_restart (pf);clear_bit (ICE_DOWN , pf->state);
@@ -5169,6 +5172,7 @@ void ice_unload(struct ice_pf *pf)devl_assert_locked (priv_to_devlink (pf)); +ice_unplug_aux_dev (pf);ice_deinit_rdma (pf);ice_deinit_features (pf);ice_tc_indir_block_unregister (vsi);
@@ -5595,6 +5599,7 @@ static int ice_suspend(struct device *dev)*/ disabled =ice_service_task_stop (pf); +ice_unplug_aux_dev (pf);ice_deinit_rdma (pf);/* Already suspended?, then there is nothing to do */
@@ -7803,7 +7808,7 @@ static void ice_rebuild(struct ice_pf *pf, enum ice_reset_req reset_type)ice_health_clear (pf); -ice_plug_aux_dev (pf); +ice_rdma_finalize_setup (pf);if (ice_is_feature_supported (pf,ICE_F_SRIOV_LAG ))ice_lag_rebuild (pf);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 196/311 ice: fix crash in ethtool offline loopback test
In reply to Sasha LevinFrom: Michal Schmidt <[email protected]>
[ Upstream commit a9c354e656597aededa027d63d2ff0973f6b033f ]
Since the conversion of ice to page pool, the ethtool loopback test crashes:
BUG: kernel NULL pointer dereference, address: 000000000000000c #PF: supervisor write access in kernel mode #PF: error_code(0x0002) - not-present page PGD 1100f1067 P4D 0 Oops: Oops: 0002 [#1] SMP NOPTI CPU: 23 UID: 0 PID: 5904 Comm: ethtool Kdump: loaded Not tainted 6.19.0-0.rc7.260128g1f97d9dcf5364.49.eln154.x86_64 #1 PREEMPT(lazy) Hardware name: [...] RIP: 0010:ice_alloc_rx_bufs+0x1cd/0x310 [ice] Code: 83 6c 24 30 01 66 41 89 47 08 0f 84 c0 00 00 00 41 0f b7 dc 48 8b 44 24 18 48 c1 e3 04 41 bb 00 10 00 00 48 8d 2c 18 8b 04 24 <89> 45 0c 41 8b 4d 00 49 d3 e3 44 3b 5c 24 24 0f 83 ac fe ff ff 44 RSP: 0018:ff7894738aa1f768 EFLAGS: 00010246 RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: 0000000000000700 RDI: 0000000000000000 RBP: 0000000000000000 R08: ff16dcae79880200 R09: 0000000000000019 R10: 0000000000000001 R11: 0000000000001000 R12: 0000000000000000 R13: 0000000000000000 R14: 0000000000000000 R15: ff16dcae6c670000 FS: 00007fcf428850c0(0000) GS:ff16dcb149710000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 000000000000000c CR3: 0000000121227005 CR4: 0000000000773ef0 PKRU: 55555554 Call Trace: <TASK> ice_vsi_cfg_rxq+0xca/0x460 [ice] ice_vsi_cfg_rxqs+0x54/0x70 [ice] ice_loopback_test+0xa9/0x520 [ice] ice_self_test+0x1b9/0x280 [ice] ethtool_self_test+0xe5/0x200 __dev_ethtool+0x1106/0x1a90 dev_ethtool+0xbe/0x1a0 dev_ioctl+0x258/0x4c0 sock_do_ioctl+0xe3/0x130 __x64_sys_ioctl+0xb9/0x100 do_syscall_64+0x7c/0x700 entry_SYSCALL_64_after_hwframe+0x76/0x7e [...]
It crashes because we have not initialized libeth for the rx ring.
Fix it by treating ICE_VSI_LB VSIs slightly more like normal PF VSIs and letting them have a q_vector. It's just a dummy, because the loopback test does not use interrupts, but it contains a napi struct that can be passed to libeth_rx_fq_create() called from ice_vsi_cfg_rxq() -> ice_rxq_pp_create().
--- drivers/net/ethernet/intel/ice/ice_base.c | 5 ++++- drivers/net/ethernet/intel/ice/ice_ethtool.c | 4 ++++ drivers/net/ethernet/intel/ice/ice_lib.c | 15 ++++++++++----- 3 files changed, 18 insertions(+), 6 deletions(-)
drivers/net/ethernet/intel/ice/ice_base.c +4 -1
@@ -124,6 +124,8 @@ static int ice_vsi_alloc_q_vector(struct ice_vsi *vsi, u16 v_idx)if (vsi->type ==ICE_VSI_VF ) {ice_calc_vf_reg_idx (vsi->vf, q_vector); goto out; + }else if (vsi->type ==ICE_VSI_LB ) { + goto skip_alloc; }else if (vsi->type ==ICE_VSI_CTRL && vsi->vf) {struct ice_vsi *ctrl_vsi =ice_get_vf_ctrl_vsi (pf, vsi);
@@ -662,7 +664,8 @@ static int ice_vsi_cfg_rxq(struct ice_rx_ring *ring)u32 rx_buf_len;int err; -if (ring->vsi->type ==ICE_VSI_PF || ring->vsi->type ==ICE_VSI_SF ) { +if (ring->vsi->type ==ICE_VSI_PF || ring->vsi->type ==ICE_VSI_SF || + ring->vsi->type ==ICE_VSI_LB ) {if (!xdp_rxq_info_is_reg (&ring->xdp_rxq)) { err = __xdp_rxq_info_reg(&ring->xdp_rxq, ring->netdev, ring->q_index,
drivers/net/ethernet/intel/ice/ice_ethtool.c +4 -0
@@ -1289,6 +1289,10 @@ static u64 ice_loopback_test(struct net_device *netdev) test_vsi->netdev = netdev; tx_ring = test_vsi->tx_rings[0 ]; rx_ring = test_vsi->rx_rings[0 ]; +/* Dummy q_vector and napi. Fill the minimum required for +* ice_rxq_pp_create(). +*/ + rx_ring->q_vector->napi.dev = netdev;if (ice_lbtest_prepare_rings (test_vsi)) { ret =2 ;
drivers/net/ethernet/intel/ice/ice_lib.c +10 -5
@@ -107,10 +107,6 @@ static int ice_vsi_alloc_arrays(struct ice_vsi *vsi)if (!vsi->rxq_map) goto err_rxq_map; -/* There is no need to allocate q_vectors for a loopback VSI. */ -if (vsi->type ==ICE_VSI_LB ) -return 0 ; -/* allocate memory for q_vector pointers */ vsi->q_vectors =devm_kcalloc (dev, vsi->num_q_vectors,sizeof (*vsi->q_vectors),GFP_KERNEL );
@@ -239,6 +235,8 @@ static void ice_vsi_set_num_qs(struct ice_vsi *vsi)case ICE_VSI_LB : vsi->alloc_txq =1 ; vsi->alloc_rxq =1 ; +/* A dummy q_vector, no actual IRQ. */ + vsi->num_q_vectors =1 ;break ;default :dev_warn (ice_pf_to_dev (pf),"Unknown VSI type %d\n" , vsi_type);
@@ -2424,14 +2422,21 @@ static int ice_vsi_cfg_def(struct ice_vsi *vsi) }break ;case ICE_VSI_LB : - ret =ice_vsi_alloc_rings (vsi); + ret =ice_vsi_alloc_q_vectors (vsi);if (ret) goto unroll_vsi_init; + ret =ice_vsi_alloc_rings (vsi); +if (ret) + goto unroll_alloc_q_vector; + ret =ice_vsi_alloc_ring_stats (vsi);if (ret) goto unroll_vector_base; +/* Simply map the dummy q_vector to the only rx_ring */ + vsi->rx_rings[0 ]->q_vector = vsi->q_vectors[0 ]; +break ;default :/* clean up the resources and exit */
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 197/311 i40e: Fix preempt count leak in napi poll tracepoint
In reply to Sasha LevinFrom: Thomas Gleixner <[email protected]>
[ Upstream commit 4b3d54a85bd37ebf2d9836f0d0de775c0ff21af9 ]
Using get_cpu() in the tracepoint assignment causes an obvious preempt count leak because nothing invokes put_cpu() to undo it:
softirq: huh, entered softirq 3 NET_RX with preempt_count 00000100, exited with 00000101?
This clearly has seen a lot of testing in the last 3+ years...
Use smp_processor_id() instead.
--- drivers/net/ethernet/intel/i40e/i40e_trace.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/intel/i40e/i40e_trace.h +1 -1
@@ -88,7 +88,7 @@ TRACE_EVENT(i40e_napi_poll, __entry->rx_clean_complete = rx_clean_complete; __entry->tx_clean_complete = tx_clean_complete; __entry->irq_num = q->irq_num; - __entry->curr_cpu =get_cpu (); + __entry->curr_cpu =smp_processor_id ();__assign_str (qname);__assign_str (dev_name); __assign_bitmask(irq_affinity, cpumask_bits(&q->affinity_mask),
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 198/311 e1000e: clear DPG_EN after reset to avoid autonomous power-gating
In reply to Sasha LevinFrom: Vitaly Lifshits <[email protected]>
[ Upstream commit 0942fc6d324eb9c6b16187b2aa994c0823557f06 ]
Panther Lake systems introduced an autonomous power gating feature for the integrated Gigabit Ethernet in shutdown state (S5) state. As part of it, the reset value of DPG_EN bit was changed to 1. Clear this bit after performing hardware reset to avoid errors such as Tx/Rx hangs, or packet loss/corruption.
--- drivers/net/ethernet/intel/e1000e/defines.h | 1 + drivers/net/ethernet/intel/e1000e/ich8lan.c | 9 +++++++++ 2 files changed, 10 insertions(+)
drivers/net/ethernet/intel/e1000e/defines.h +1 -0
@@ -33,6 +33,7 @@/* Extended Device Control */ #define E1000_CTRL_EXT_LPCD 0x00000004/* LCD Power Cycle Done */ +#define E1000_CTRL_EXT_DPG_EN 0x00000008/* Dynamic Power Gating Enable */ #define E1000_CTRL_EXT_SDP3_DATA 0x00000080/* Value of SW Definable Pin 3 */ #define E1000_CTRL_EXT_FORCE_SMBUS 0x00000800/* Force SMBus mode */ #define E1000_CTRL_EXT_EE_RST 0x00002000/* Reinitialize from EEPROM */
drivers/net/ethernet/intel/e1000e/ich8lan.c +9 -0
@@ -4932,6 +4932,15 @@ static s32 e1000_reset_hw_ich8lan(struct e1000_hw *hw) reg |=E1000_KABGTXD_BGSQLBIAS ;ew32 (KABGTXD , reg); +/* The hardware reset value of the DPG_EN bit is 1. +* Clear DPG_EN to prevent unexpected autonomous power gating. +*/ +if (hw->mac.type >= e1000_pch_ptp) { + reg =er32 (CTRL_EXT ); + reg &= ~E1000_CTRL_EXT_DPG_EN ; +ew32 (CTRL_EXT , reg); + } +return 0 ; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 199/311 drm/solomon: Fix page start when updating rectangle in page addressing mode
In reply to Sasha LevinFrom: Francesco Lavra <[email protected]>
[ Upstream commit 36d9579fed6c9429aa172f77bd28c58696ce8e2b ]
In page addressing mode, the pixel values of a dirty rectangle must be sent to the display controller one page at a time. The range of pages corresponding to a given rectangle is being incorrectly calculated as if the Y value of the top left coordinate of the rectangle was 0. This can result in rectangle updates being displayed on wrong parts of the screen.
Fix the above issue by consolidating the start page calculation in a single place at the beginning of the update_rect function, and using the calculated value for all addressing modes.
--- drivers/gpu/drm/solomon/ssd130x.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-)
drivers/gpu/drm/solomon/ssd130x.c +2 -4
@@ -737,6 +737,7 @@ static int ssd130x_update_rect(struct ssd130x_device *ssd130x,unsigned int height =drm_rect_height (rect);unsigned int line_length =DIV_ROUND_UP (width,8 );unsigned int page_height =SSD130X_PAGE_HEIGHT ; +u8 page_start = ssd130x->page_offset + y / page_height;unsigned int pages =DIV_ROUND_UP (height, page_height);struct drm_device *drm = &ssd130x->drm;u32 array_idx =0 ;
@@ -774,14 +775,11 @@ static int ssd130x_update_rect(struct ssd130x_device *ssd130x,*/ if (!ssd130x->page_address_mode) { -u8 page_start; -/* Set address range for horizontal addressing mode */ ret =ssd130x_set_col_range (ssd130x, ssd130x->col_offset + x, width);if (ret <0 )return ret; - page_start = ssd130x->page_offset + y / page_height; ret =ssd130x_set_page_range (ssd130x, page_start, pages);if (ret <0 )return ret;
@@ -813,7 +811,7 @@ static int ssd130x_update_rect(struct ssd130x_device *ssd130x,*/ if (ssd130x->page_address_mode) { ret =ssd130x_set_page_pos (ssd130x, - ssd130x->page_offset + i, + page_start + i, ssd130x->col_offset + x);if (ret <0 )return ret;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 200/311 netfs: Fix unbuffered/DIO writes to dispatch subrequests in strict sequence
In reply to Sasha LevinFrom: David Howells <[email protected]>
[ Upstream commit a0b4c7a49137ed21279f354eb59f49ddae8dffc2 ]
Fix netfslib such that when it's making an unbuffered or DIO write, to make sure that it sends each subrequest strictly sequentially, waiting till the previous one is 'committed' before sending the next so that we don't have pieces landing out of order and potentially leaving a hole if an error occurs (ENOSPC for example).
This is done by copying in just those bits of issuing, collecting and retrying subrequests that are necessary to do one subrequest at a time. Retrying, in particular, is simpler because if the current subrequest needs retrying, the source iterator can just be copied again and the subrequest prepped and issued again without needing to be concerned about whether it needs merging with the previous or next in the sequence.
Note that the issuing loop waits for a subrequest to complete right after issuing it, but this wait could be moved elsewhere allowing preparatory steps to be performed whilst the subrequest is in progress. In particular, once content encryption is available in netfslib, that could be done whilst waiting, as could cleanup of buffers that have been completed.
--- fs/netfs/direct_write.c | 228 ++++++++++++++++++++++++++++++++--- fs/netfs/internal.h | 4 +- fs/netfs/write_collect.c | 21 ---- fs/netfs/write_issue.c | 41 +------ include/trace/events/netfs.h | 4 +- 5 files changed, 221 insertions(+), 77 deletions(-)
fs/netfs/direct_write.c +212 -16
@@ -9,6 +9,202 @@#include <linux/uio.h> #include "internal.h" +/* +* Perform the cleanup rituals after an unbuffered write is complete. +*/ +static void netfs_unbuffered_write_done (struct netfs_io_request *wreq) +{ +struct netfs_inode *ictx =netfs_inode (wreq->inode); + +_enter ("R=%x" , wreq->debug_id); + +/* Okay, declare that all I/O is complete. */ +trace_netfs_rreq (wreq, netfs_rreq_trace_write_done); + +if (!wreq->error) +netfs_update_i_size (ictx, &ictx->inode, wreq->start, wreq->transferred); + +if (wreq->origin ==NETFS_DIO_WRITE && + wreq->mapping->nrpages) { +/* mmap may have got underfoot and we may now have folios +* locally covering the region we just wrote. Attempt to +* discard the folios, but leave in place any modified locally. +* ->write_iter() is prevented from interfering by the DIO +* counter. +*/ +pgoff_t first = wreq->start >>PAGE_SHIFT ; +pgoff_t last = (wreq->start + wreq->transferred -1 ) >>PAGE_SHIFT ; + +invalidate_inode_pages2_range (wreq->mapping, first, last); + } + +if (wreq->origin ==NETFS_DIO_WRITE ) +inode_dio_end (wreq->inode); + +_debug ("finished" ); +netfs_wake_rreq_flag (wreq,NETFS_RREQ_IN_PROGRESS , netfs_rreq_trace_wake_ip); +/* As we cleared NETFS_RREQ_IN_PROGRESS, we acquired its ref. */ + +if (wreq->iocb) { +size_t written =umin (wreq->transferred, wreq->len); + + wreq->iocb->ki_pos += written; +if (wreq->iocb->ki_complete) { +trace_netfs_rreq (wreq, netfs_rreq_trace_ki_complete); + wreq->iocb->ki_complete (wreq->iocb, wreq->error ?: written); + } + wreq->iocb =VFS_PTR_POISON ; + } + +netfs_clear_subrequests (wreq); +} + +/* +* Collect the subrequest results of unbuffered write subrequests. +*/ +static void netfs_unbuffered_write_collect (struct netfs_io_request *wreq, +struct netfs_io_stream *stream, +struct netfs_io_subrequest *subreq) +{ +trace_netfs_collect_sreq (wreq, subreq); + +spin_lock (&wreq->lock); +list_del_init (&subreq->rreq_link); +spin_unlock (&wreq->lock); + + wreq->transferred += subreq->transferred; +iov_iter_advance (&wreq->buffer.iter, subreq->transferred); + + stream->collected_to = subreq->start + subreq->transferred; + wreq->collected_to = stream->collected_to; +netfs_put_subrequest (subreq, netfs_sreq_trace_put_done); + +trace_netfs_collect_stream (wreq, stream); +trace_netfs_collect_state (wreq, wreq->collected_to,0 ); +} + +/* +* Write data to the server without going through the pagecache and without +* writing it to the local cache. We dispatch the subrequests serially and +* wait for each to complete before dispatching the next, lest we leave a gap +* in the data written due to a failure such as ENOSPC. We could, however +* attempt to do preparation such as content encryption for the next subreq +* whilst the current is in progress. +*/ +static int netfs_unbuffered_write (struct netfs_io_request *wreq) +{ +struct netfs_io_subrequest *subreq =NULL ; +struct netfs_io_stream *stream = &wreq->io_streams[0 ]; +int ret; + +_enter ("%llx" , wreq->len); + +if (wreq->origin ==NETFS_DIO_WRITE ) +inode_dio_begin (wreq->inode); + + stream->collected_to = wreq->start; + +for (;;) { +bool retry = false; + +if (!subreq) { +netfs_prepare_write (wreq, stream, wreq->start + wreq->transferred); + subreq = stream->construct; + stream->construct =NULL ; + stream->front =NULL ; + } + +/* Check if (re-)preparation failed. */ +if (unlikely (test_bit (NETFS_SREQ_FAILED , &subreq->flags))) { +netfs_write_subrequest_terminated (subreq, subreq->error); + wreq->error = subreq->error; +break ; + } + +iov_iter_truncate (&subreq->io_iter, wreq->len - wreq->transferred); +if (!iov_iter_count (&subreq->io_iter)) +break ; + + subreq->len =netfs_limit_iter (&subreq->io_iter,0 , + stream->sreq_max_len, + stream->sreq_max_segs); +iov_iter_truncate (&subreq->io_iter, subreq->len); + stream->submit_extendable_to = subreq->len; + +trace_netfs_sreq (subreq, netfs_sreq_trace_submit); + stream->issue_write (subreq); + +/* Async, need to wait. */ +netfs_wait_for_in_progress_stream (wreq, stream); + +if (test_bit (NETFS_SREQ_NEED_RETRY , &subreq->flags)) { + retry = true; + }else if (test_bit (NETFS_SREQ_FAILED , &subreq->flags)) { + ret = subreq->error; + wreq->error = ret; +netfs_see_subrequest (subreq, netfs_sreq_trace_see_failed); + subreq =NULL ; +break ; + } + ret =0 ; + +if (!retry) { +netfs_unbuffered_write_collect (wreq, stream, subreq); + subreq =NULL ; +if (wreq->transferred >= wreq->len) +break ; +if (!wreq->iocb &&signal_pending (current)) { + ret = wreq->transferred ? -EINTR : -ERESTARTSYS ; +trace_netfs_rreq (wreq, netfs_rreq_trace_intr); +break ; + } +continue ; + } + +/* We need to retry the last subrequest, so first reset the +* iterator, taking into account what, if anything, we managed +* to transfer. +*/ + subreq->error = -EAGAIN ; +trace_netfs_sreq (subreq, netfs_sreq_trace_retry); +if (subreq->transferred >0 ) +iov_iter_advance (&wreq->buffer.iter, subreq->transferred); + +if (stream->source ==NETFS_UPLOAD_TO_SERVER && + wreq->netfs_ops->retry_request) + wreq->netfs_ops->retry_request (wreq, stream); + +__clear_bit (NETFS_SREQ_NEED_RETRY , &subreq->flags); +__clear_bit (NETFS_SREQ_BOUNDARY , &subreq->flags); +__clear_bit (NETFS_SREQ_FAILED , &subreq->flags); + subreq->io_iter = wreq->buffer.iter; + subreq->start = wreq->start + wreq->transferred; + subreq->len = wreq->len - wreq->transferred; + subreq->transferred =0 ; + subreq->retry_count +=1 ; + stream->sreq_max_len =UINT_MAX ; + stream->sreq_max_segs =INT_MAX ; + +netfs_get_subrequest (subreq, netfs_sreq_trace_get_resubmit); + stream->prepare_write (subreq); + +__set_bit (NETFS_SREQ_IN_PROGRESS , &subreq->flags); +netfs_stat (&netfs_n_wh_retry_write_subreq); + } + +netfs_unbuffered_write_done (wreq); +_leave (" = %d" , ret); +return ret; +} + +static void netfs_unbuffered_write_async (struct work_struct *work) +{ + struct netfs_io_request *wreq = container_of(work, struct netfs_io_request, work); + +netfs_unbuffered_write (wreq); +netfs_put_request (wreq, netfs_rreq_trace_put_complete); +} +/* * Perform an unbuffered write where we may have to do an RMW operation on an * encrypted file. This can also be used for direct I/O writes.
@@ -70,35 +266,35 @@ ssize_t netfs_unbuffered_write_iter_locked(struct kiocb *iocb, struct iov_iter **/ wreq->buffer.iter = *iter; } + + wreq->len =iov_iter_count (&wreq->buffer.iter); }__set_bit (NETFS_RREQ_USE_IO_ITER , &wreq->flags); -if (async) -__set_bit (NETFS_RREQ_OFFLOAD_COLLECTION , &wreq->flags);/* Copy the data into the bounce buffer and encrypt it. */ // TODO /* Dispatch the write. */ __set_bit (NETFS_RREQ_UPLOAD_TO_SERVER , &wreq->flags); -if (async) - wreq->iocb = iocb; - wreq->len =iov_iter_count (&wreq->buffer.iter); - ret =netfs_unbuffered_write (wreq,is_sync_kiocb (iocb), wreq->len); -if (ret <0 ) { -_debug ("begin = %zd" , ret); - goto out; - } -if (!async) { - ret =netfs_wait_for_write (wreq); -if (ret >0 ) - iocb->ki_pos += ret; - }else { +if (async) { +INIT_WORK (&wreq->work, netfs_unbuffered_write_async); + wreq->iocb = iocb; +queue_work (system_dfl_wq, &wreq->work); ret = -EIOCBQUEUED ; + }else { + ret =netfs_unbuffered_write (wreq); +if (ret <0 ) { +_debug ("begin = %zd" , ret); + }else { + iocb->ki_pos += wreq->transferred; + ret = wreq->transferred ?: wreq->error; + } + +netfs_put_request (wreq, netfs_rreq_trace_put_complete); } -out:netfs_put_request (wreq, netfs_rreq_trace_put_return);return ret;
fs/netfs/internal.h +3 -1
@@ -198,6 +198,9 @@ struct netfs_io_request *netfs_create_write_req(struct address_space *mapping,struct file *file, loff_t start, enum netfs_io_origin origin); +void netfs_prepare_write (struct netfs_io_request *wreq, +struct netfs_io_stream *stream, +loff_t start);void netfs_reissue_write (struct netfs_io_stream *stream,struct netfs_io_subrequest *subreq,struct iov_iter *source);
@@ -212,7 +215,6 @@ int netfs_advance_writethrough(struct netfs_io_request *wreq, struct writeback_c struct folio **writethrough_cache);ssize_t netfs_end_writethrough (struct netfs_io_request *wreq,struct writeback_control *wbc,struct folio *writethrough_cache); -int netfs_unbuffered_write (struct netfs_io_request *wreq,bool may_wait,size_t len);/* * write_retry.c
fs/netfs/write_collect.c +0 -21
@@ -399,27 +399,6 @@ bool netfs_write_collection(struct netfs_io_request *wreq) ictx->ops->invalidate_cache (wreq); } -if ((wreq->origin ==NETFS_UNBUFFERED_WRITE || - wreq->origin ==NETFS_DIO_WRITE ) && - !wreq->error) -netfs_update_i_size (ictx, &ictx->inode, wreq->start, wreq->transferred); - -if (wreq->origin ==NETFS_DIO_WRITE && - wreq->mapping->nrpages) { -/* mmap may have got underfoot and we may now have folios -* locally covering the region we just wrote. Attempt to -* discard the folios, but leave in place any modified locally. -* ->write_iter() is prevented from interfering by the DIO -* counter. -*/ -pgoff_t first = wreq->start >>PAGE_SHIFT ; -pgoff_t last = (wreq->start + wreq->transferred -1 ) >>PAGE_SHIFT ; -invalidate_inode_pages2_range (wreq->mapping, first, last); - } - -if (wreq->origin ==NETFS_DIO_WRITE ) -inode_dio_end (wreq->inode); -_debug ("finished" );netfs_wake_rreq_flag (wreq,NETFS_RREQ_IN_PROGRESS , netfs_rreq_trace_wake_ip);/* As we cleared NETFS_RREQ_IN_PROGRESS, we acquired its ref. */
fs/netfs/write_issue.c +3 -38
53 lines not on this page · raw message
include/trace/events/netfs.h +3 -1
@@ -57,6 +57,7 @@ EM(netfs_rreq_trace_done, "DONE ") \ EM(netfs_rreq_trace_end_copy_to_cache, "END-C2C") \ EM(netfs_rreq_trace_free, "FREE ") \ + EM(netfs_rreq_trace_intr, "INTR ") \ EM(netfs_rreq_trace_ki_complete, "KI-CMPL") \ EM(netfs_rreq_trace_recollect, "RECLLCT") \EM (netfs_rreq_trace_redirty,"REDIRTY" ) \
@@ -169,7 +170,8 @@ EM(netfs_sreq_trace_put_oom, "PUT OOM ") \ EM(netfs_sreq_trace_put_wip, "PUT WIP ") \ EM(netfs_sreq_trace_put_work, "PUT WORK ") \ - E_(netfs_sreq_trace_put_terminated, "PUT TERM ") + EM(netfs_sreq_trace_put_terminated, "PUT TERM ") \ + E_(netfs_sreq_trace_see_failed, "SEE FAILED ")#define netfs_folio_traces \ EM(netfs_folio_is_uptodate, "mod-uptodate") \
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 201/311 nvmet-fcloop: Check remoteport port_state before calling done callback
In reply to Sasha LevinFrom: Justin Tee <[email protected]>
[ Upstream commit dd677d0598387ea623820ab2bd0e029c377445a3 ]
In nvme_fc_handle_ls_rqst_work, the lsrsp->done callback is only set when remoteport->port_state is FC_OBJSTATE_ONLINE. Otherwise, the nvme_fc_xmt_ls_rsp's LLDD call to lport->ops->xmt_ls_rsp is expected to fail and the nvme-fc transport layer itself will directly call nvme_fc_xmt_ls_rsp_free instead of relying on LLDD's done callback to free the lsrsp resources.
Update the fcloop_t2h_xmt_ls_rsp routine to check remoteport->port_state. If online, then lsrsp->done callback will free the lsrsp. Else, return -ENODEV to signal the nvme-fc transport to handle freeing lsrsp.
--- drivers/nvme/target/fcloop.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-)
drivers/nvme/target/fcloop.c +11 -4
@@ -491,6 +491,7 @@ fcloop_t2h_xmt_ls_rsp(struct nvme_fc_local_port *localport,struct fcloop_rport *rport = remoteport->private;struct nvmet_fc_target_port *targetport = rport->targetport;struct fcloop_tport *tport; +int ret =0 ;if (!targetport) {/*
@@ -500,12 +501,18 @@ fcloop_t2h_xmt_ls_rsp(struct nvme_fc_local_port *localport,* We end up here from delete association exchange: * nvmet_fc_xmt_disconnect_assoc sends an async request. * -* Return success because this is what LLDDs do; silently -* drop the response. +* Return success when remoteport is still online because this +* is what LLDDs do and silently drop the response. Otherwise, +* return with error to signal upper layer to perform the lsrsp +* resource cleanup. */ - lsrsp->done (lsrsp); +if (remoteport->port_state ==FC_OBJSTATE_ONLINE ) + lsrsp->done (lsrsp); +else + ret = -ENODEV ; +kmem_cache_free (lsreq_cache, tls_req); -return 0 ; +return ret; } memcpy(lsreq->rspaddr, lsrsp->rspbuf,
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 202/311 net: annotate data-races around sk->sk_{data_ready,write_space}
In reply to Sasha LevinFrom: Eric Dumazet <[email protected]>
[ Upstream commit 2ef2b20cf4e04ac8a6ba68493f8780776ff84300 ]
skmsg (and probably other layers) are changing these pointers while other cpus might read them concurrently.
Add corresponding READ_ONCE()/WRITE_ONCE() annotations for UDP, TCP and AF_UNIX.
--- net/core/skmsg.c | 14 +++++++------- net/ipv4/tcp.c | 4 ++-- net/ipv4/tcp_bpf.c | 2 +- net/ipv4/tcp_input.c | 14 ++++++++------ net/ipv4/tcp_minisocks.c | 2 +- net/ipv4/udp.c | 2 +- net/ipv4/udp_bpf.c | 2 +- net/unix/af_unix.c | 8 ++++---- 8 files changed, 25 insertions(+), 23 deletions(-)
net/core/skmsg.c +7 -7
@@ -1205,8 +1205,8 @@ void sk_psock_start_strp(struct sock *sk, struct sk_psock *psock)return ; psock->saved_data_ready = sk->sk_data_ready; - sk->sk_data_ready = sk_psock_strp_data_ready; - sk->sk_write_space = sk_psock_write_space; +WRITE_ONCE (sk->sk_data_ready, sk_psock_strp_data_ready); +WRITE_ONCE (sk->sk_write_space, sk_psock_write_space); }void sk_psock_stop_strp (struct sock *sk,struct sk_psock *psock)
@@ -1216,8 +1216,8 @@ void sk_psock_stop_strp(struct sock *sk, struct sk_psock *psock)if (!psock->saved_data_ready)return ; - sk->sk_data_ready = psock->saved_data_ready; - psock->saved_data_ready =NULL ; +WRITE_ONCE (sk->sk_data_ready, psock->saved_data_ready); +WRITE_ONCE (psock->saved_data_ready,NULL );strp_stop (&psock->strp); }
@@ -1296,8 +1296,8 @@ void sk_psock_start_verdict(struct sock *sk, struct sk_psock *psock)return ; psock->saved_data_ready = sk->sk_data_ready; - sk->sk_data_ready = sk_psock_verdict_data_ready; - sk->sk_write_space = sk_psock_write_space; +WRITE_ONCE (sk->sk_data_ready, sk_psock_verdict_data_ready); +WRITE_ONCE (sk->sk_write_space, sk_psock_write_space); }void sk_psock_stop_verdict (struct sock *sk,struct sk_psock *psock)
@@ -1308,6 +1308,6 @@ void sk_psock_stop_verdict(struct sock *sk, struct sk_psock *psock)if (!psock->saved_data_ready)return ; - sk->sk_data_ready = psock->saved_data_ready; +WRITE_ONCE (sk->sk_data_ready, psock->saved_data_ready); psock->saved_data_ready =NULL ; }
net/ipv4/tcp.c +2 -2
@@ -1397,7 +1397,7 @@ int tcp_sendmsg_locked(struct sock *sk, struct msghdr *msg, size_t size) err =sk_stream_error (sk, flags, err);/* make sure we wake any epoll edge trigger waiter */ if (unlikely (tcp_rtx_and_write_queues_empty (sk) && err == -EAGAIN )) { - sk->sk_write_space (sk); +READ_ONCE (sk->sk_write_space)(sk);tcp_chrono_stop (sk,TCP_CHRONO_SNDBUF_LIMITED ); }if (binding)
@@ -4131,7 +4131,7 @@ int do_tcp_setsockopt(struct sock *sk, int level, int optname,break ;case TCP_NOTSENT_LOWAT :WRITE_ONCE (tp->notsent_lowat, val); - sk->sk_write_space (sk); +READ_ONCE (sk->sk_write_space)(sk);break ;case TCP_INQ :if (val >1 || val <0 )
net/ipv4/tcp_bpf.c +1 -1
@@ -725,7 +725,7 @@ int tcp_bpf_update_proto(struct sock *sk, struct sk_psock *psock, bool restore)WRITE_ONCE (sk->sk_prot->unhash, psock->saved_unhash);tcp_update_ulp (sk, psock->sk_proto, psock->saved_write_space); }else { - sk->sk_write_space = psock->saved_write_space; +WRITE_ONCE (sk->sk_write_space, psock->saved_write_space);/* Pairs with lockless read in sk_clone_lock() */ sock_replace_proto (sk, psock->sk_proto); }
net/ipv4/tcp_input.c +8 -6
@@ -5167,7 +5167,7 @@ static void tcp_data_queue_ofo(struct sock *sk, struct sk_buff *skb)if (unlikely (tcp_try_rmem_schedule (sk, skb, skb->truesize))) {NET_INC_STATS (sock_net (sk),LINUX_MIB_TCPOFODROP ); - sk->sk_data_ready (sk); +READ_ONCE (sk->sk_data_ready)(sk);tcp_drop_reason (sk, skb,SKB_DROP_REASON_PROTO_MEM );return ; }
@@ -5377,7 +5377,7 @@ int tcp_send_rcvq(struct sock *sk, struct msghdr *msg, size_t size)void tcp_data_ready (struct sock *sk) {if (tcp_epollin_ready (sk, sk->sk_rcvlowat) ||sock_flag (sk,SOCK_DONE )) - sk->sk_data_ready (sk); +READ_ONCE (sk->sk_data_ready)(sk); }static void tcp_data_queue (struct sock *sk,struct sk_buff *skb)
@@ -5433,7 +5433,7 @@ static void tcp_data_queue(struct sock *sk, struct sk_buff *skb)inet_csk (sk)->icsk_ack.pending |= (ICSK_ACK_NOMEM |ICSK_ACK_NOW );inet_csk_schedule_ack (sk); - sk->sk_data_ready (sk); +READ_ONCE (sk->sk_data_ready)(sk);if (skb_queue_len (&sk->sk_receive_queue) && skb->len) { reason =SKB_DROP_REASON_PROTO_MEM ;
@@ -5875,7 +5875,9 @@ static void tcp_new_space(struct sock *sk) tp->snd_cwnd_stamp = tcp_jiffies32; } -INDIRECT_CALL_1 (sk->sk_write_space, sk_stream_write_space, sk); +INDIRECT_CALL_1 (READ_ONCE (sk->sk_write_space), + sk_stream_write_space, + sk); }/* Caller made space either from:
@@ -6091,7 +6093,7 @@ static void tcp_urg(struct sock *sk, struct sk_buff *skb, const struct tcphdr *tBUG ();WRITE_ONCE (tp->urg_data,TCP_URG_VALID | tmp);if (!sock_flag (sk,SOCK_DEAD )) - sk->sk_data_ready (sk); +READ_ONCE (sk->sk_data_ready)(sk); } } }
@@ -7557,7 +7559,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops,sock_put (fastopen_sk); goto drop_and_free; } - sk->sk_data_ready (sk); +READ_ONCE (sk->sk_data_ready)(sk);bh_unlock_sock (fastopen_sk);sock_put (fastopen_sk); }else {
net/ipv4/tcp_minisocks.c +1 -1
@@ -988,7 +988,7 @@ enum skb_drop_reason tcp_child_process(struct sock *parent, struct sock *child, reason =tcp_rcv_state_process (child, skb);/* Wakeup parent, send SIGIO */ if (state ==TCP_SYN_RECV && child->sk_state != state) - parent->sk_data_ready (parent); +READ_ONCE (parent->sk_data_ready)(parent); }else {/* Alas, it is possible again, because we do lookup * in main socket hash table and lock on listening
net/ipv4/udp.c +1 -1
@@ -1786,7 +1786,7 @@ int __udp_enqueue_schedule_skb(struct sock *sk, struct sk_buff *skb)* using prepare_to_wait_exclusive(). */ while (nb) { -INDIRECT_CALL_1 (sk->sk_data_ready, +INDIRECT_CALL_1 (READ_ONCE (sk->sk_data_ready), sock_def_readable, sk); nb--; }
net/ipv4/udp_bpf.c +1 -1
@@ -158,7 +158,7 @@ int udp_bpf_update_proto(struct sock *sk, struct sk_psock *psock, bool restore)int family = sk->sk_family ==AF_INET ?UDP_BPF_IPV4 :UDP_BPF_IPV6 ;if (restore) { - sk->sk_write_space = psock->saved_write_space; +WRITE_ONCE (sk->sk_write_space, psock->saved_write_space);sock_replace_proto (sk, psock->sk_proto);return 0 ; }
net/unix/af_unix.c +4 -4
@@ -1785,7 +1785,7 @@ static int unix_stream_connect(struct socket *sock, struct sockaddr_unsized *uad__skb_queue_tail (&other->sk_receive_queue, skb);spin_unlock (&other->sk_receive_queue.lock);unix_state_unlock (other); - other->sk_data_ready (other); +READ_ONCE (other->sk_data_ready)(other);sock_put (other);return 0 ;
@@ -2278,7 +2278,7 @@ static int unix_dgram_sendmsg(struct socket *sock, struct msghdr *msg,scm_stat_add (other, skb);skb_queue_tail (&other->sk_receive_queue, skb);unix_state_unlock (other); - other->sk_data_ready (other); +READ_ONCE (other->sk_data_ready)(other);sock_put (other);scm_destroy (&scm);return len;
@@ -2351,7 +2351,7 @@ static int queue_oob(struct sock *sk, struct msghdr *msg, struct sock *other,sk_send_sigurg (other);unix_state_unlock (other); - other->sk_data_ready (other); +READ_ONCE (other->sk_data_ready)(other);return 0 ; out_unlock:
@@ -2477,7 +2477,7 @@ static int unix_stream_sendmsg(struct socket *sock, struct msghdr *msg,spin_unlock (&other->sk_receive_queue.lock);unix_state_unlock (other); - other->sk_data_ready (other); +READ_ONCE (other->sk_data_ready)(other); sent += size; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 203/311 bridge: Check relevant per-VLAN options in VLAN range grouping
In reply to Sasha LevinFrom: Danielle Ratson <[email protected]>
[ Upstream commit 93c9475c04acad2457a7e7ea4e3ec40a6e6d94a7 ]
The br_vlan_opts_eq_range() function determines if consecutive VLANs can be grouped together in a range for compact netlink notifications. It currently checks state, tunnel info, and multicast router configuration, but misses two categories of per-VLAN options that affect the output:
1. User-visible priv_flags (neigh_suppress, mcast_enabled)
2. Port multicast context (mcast_max_groups, mcast_n_groups)
When VLANs have different settings for these options, they are incorrectly grouped into ranges, causing netlink notifications to report only one VLAN's settings for the entire range.
Fix by checking priv_flags equality, but only for flags that affect netlink output (BR_VLFLAG_NEIGH_SUPPRESS_ENABLED and BR_VLFLAG_MCAST_ENABLED), and comparing multicast context (mcast_max_groups and mcast_n_groups).
Example showing the bugs before the fix:
$ bridge vlan set vid 10 dev dummy1 neigh_suppress on
$ bridge vlan set vid 11 dev dummy1 neigh_suppress off
$ bridge -d vlan show dev dummy1
port vlan-id
dummy1 10-11
... neigh_suppress on$ bridge vlan set vid 10 dev dummy1 mcast_max_groups 100
$ bridge vlan set vid 11 dev dummy1 mcast_max_groups 200
$ bridge -d vlan show dev dummy1
port vlan-id
dummy1 10-11
... mcast_max_groups 100After the fix, VLANs 10 and 11 are shown as separate entries with their correct individual settings.
--- net/bridge/br_private.h | 10 ++++++++++ net/bridge/br_vlan_options.c | 26 +++++++++++++++++++++++--- 2 files changed, 33 insertions(+), 3 deletions(-)
net/bridge/br_private.h +10 -0
@@ -1344,6 +1344,16 @@ br_multicast_ctx_options_equal(const struct net_bridge_mcast *brmctx1, true; } +static inline bool +br_multicast_port_ctx_options_equal (const struct net_bridge_mcast_port *pmctx1, +const struct net_bridge_mcast_port *pmctx2) +{ +return br_multicast_ngroups_get (pmctx1) == +br_multicast_ngroups_get (pmctx2) && +br_multicast_ngroups_get_max (pmctx1) == +br_multicast_ngroups_get_max (pmctx2); +} +static inline bool br_multicast_ctx_matches_vlan_snooping (const struct net_bridge_mcast *brmctx) {
net/bridge/br_vlan_options.c +23 -3
@@ -43,9 +43,29 @@ bool br_vlan_opts_eq_range(const struct net_bridge_vlan *v_curr,u8 range_mc_rtr =br_vlan_multicast_router (range_end);u8 curr_mc_rtr =br_vlan_multicast_router (v_curr); -return v_curr->state == range_end->state && -__vlan_tun_can_enter_range (v_curr, range_end) && - curr_mc_rtr == range_mc_rtr; +if (v_curr->state != range_end->state) +return false; + +if (!__vlan_tun_can_enter_range (v_curr, range_end)) +return false; + +if (curr_mc_rtr != range_mc_rtr) +return false; + +/* Check user-visible priv_flags that affect output */ +if ((v_curr->priv_flags ^ range_end->priv_flags) & + (BR_VLFLAG_NEIGH_SUPPRESS_ENABLED |BR_VLFLAG_MCAST_ENABLED )) +return false; + +#ifdef CONFIG_BRIDGE_IGMP_SNOOPING +if (!br_vlan_is_master (v_curr) && + !br_multicast_port_ctx_vlan_disabled (&v_curr->port_mcast_ctx) && + !br_multicast_port_ctx_options_equal (&v_curr->port_mcast_ctx, + &range_end->port_mcast_ctx)) +return false; +#endif + +return true; } bool br_vlan_opts_fill(struct sk_buff *skb, const struct net_bridge_vlan *v,
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 204/311 net: ethernet: ti: am65-cpsw-nuss/cpsw-ale: Fix multicast entry handling in ALE table
In reply to Sasha LevinFrom: Chintan Vankar <[email protected]>
[ Upstream commit be11a537224d72b906db6b98510619770298c8a4 ]
In the current implementation, flushing multicast entries in MAC mode incorrectly deletes entries for all ports instead of only the target port, disrupting multicast traffic on other ports. The cause is adding multicast entries by setting only host port bit, and not setting the MAC port bits.
Fix this by setting the MAC port's bit in the port mask while adding the multicast entry. Also fix the flush logic to preserve the host port bit during removal of MAC port and free ALE entries when mask contains only host port.
--- drivers/net/ethernet/ti/am65-cpsw-nuss.c | 2 +- drivers/net/ethernet/ti/cpsw_ale.c | 9 ++++----- 2 files changed, 5 insertions(+), 6 deletions(-)
drivers/net/ethernet/ti/am65-cpsw-nuss.c +1 -1
@@ -391,7 +391,7 @@ static void am65_cpsw_nuss_ndo_slave_set_rx_mode(struct net_device *ndev)cpsw_ale_set_allmulti (common->ale, ndev->flags &IFF_ALLMULTI , port->port_id); - port_mask =ALE_PORT_HOST ; + port_mask =BIT (port->port_id) |ALE_PORT_HOST ;/* Clear all mcast from ALE */ cpsw_ale_flush_multicast (common->ale, port_mask,-1 );
drivers/net/ethernet/ti/cpsw_ale.c +4 -5
@@ -455,14 +455,13 @@ static void cpsw_ale_flush_mcast(struct cpsw_ale *ale, u32 *ale_entry, ale->port_mask_bits);if ((mask & port_mask) ==0 )return ;/* ports dont intersect, not interested */ - mask &= ~port_mask; + mask &= (~port_mask |ALE_PORT_HOST ); -/* free if only remaining port is host port */ -if (mask) +if (mask ==0x0 || mask ==ALE_PORT_HOST ) +cpsw_ale_set_entry_type (ale_entry,ALE_TYPE_FREE ); +else cpsw_ale_set_port_mask (ale_entry, mask, ale->port_mask_bits); -else -cpsw_ale_set_entry_type (ale_entry,ALE_TYPE_FREE ); }int cpsw_ale_flush_multicast (struct cpsw_ale *ale,int port_mask,int vid)
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 205/311 nvme-multipath: fix leak on try_module_get failure
In reply to Sasha LevinFrom: Keith Busch <[email protected]>
[ Upstream commit 0f5197ea9a73a4c406c75e6d8af3a13f7f96ae89 ]
We need to fall back to the synchronous removal if we can't get a reference on the module needed for the deferred removal.
--- drivers/nvme/host/multipath.c | 12 +++++------- 1 file changed, 5 insertions(+), 7 deletions(-)
drivers/nvme/host/multipath.c +5 -7
@@ -1310,13 +1310,11 @@ void nvme_mpath_remove_disk(struct nvme_ns_head *head)if (!list_empty (&head->list)) goto out; -if (head->delayed_removal_secs) { -/* -* Ensure that no one could remove this module while the head -* remove work is pending. -*/ -if (!try_module_get (THIS_MODULE )) - goto out; +/* +* Ensure that no one could remove this module while the head +* remove work is pending. +*/ +if (head->delayed_removal_secs &&try_module_get (THIS_MODULE )) {mod_delayed_work (nvme_wq, &head->remove_work, head->delayed_removal_secs *HZ ); }else {
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 206/311 inet: annotate data-races around isk->inet_num
In reply to Sasha LevinFrom: Eric Dumazet <[email protected]>
[ Upstream commit 29252397bcc1e0a1f85e5c3bee59c325f5c26341 ]
UDP/TCP lookups are using RCU, thus isk->inet_num accesses should use READ_ONCE() and WRITE_ONCE() where needed.
--- include/net/inet6_hashtables.h | 2 +- include/net/inet_hashtables.h | 2 +- include/net/ip.h | 2 +- net/ipv4/inet_hashtables.c | 8 ++++---- net/ipv4/tcp_diag.c | 2 +- net/ipv6/inet6_hashtables.c | 3 ++- 6 files changed, 10 insertions(+), 9 deletions(-)
include/net/inet6_hashtables.h +1 -1
@@ -175,7 +175,7 @@ static inline bool inet6_match(const struct net *net, const struct sock *sk, {if (!net_eq (sock_net (sk), net) || sk->sk_family !=AF_INET6 || - sk->sk_portpair != ports || +READ_ONCE (sk->sk_portpair) != ports || !ipv6_addr_equal (&sk->sk_v6_daddr, saddr) || !ipv6_addr_equal (&sk->sk_v6_rcv_saddr, daddr))return false;
include/net/inet_hashtables.h +1 -1
@@ -345,7 +345,7 @@ static inline bool inet_match(const struct net *net, const struct sock *sk, int dif, int sdif) {if (!net_eq (sock_net (sk), net) || - sk->sk_portpair != ports || +READ_ONCE (sk->sk_portpair) != ports || sk->sk_addrpair != cookie)return false;
include/net/ip.h +1 -1
@@ -101,7 +101,7 @@ static inline void ipcm_init_sk(struct ipcm_cookie *ipcm, ipcm->oif =READ_ONCE (inet->sk.sk_bound_dev_if); ipcm->addr = inet->inet_saddr; - ipcm->protocol = inet->inet_num; + ipcm->protocol =READ_ONCE (inet->inet_num); }#define IPCB (skb) ((struct inet_skb_parm*)((skb)->cb))
net/ipv4/inet_hashtables.c +4 -4
@@ -200,7 +200,7 @@ static bool inet_bind2_bucket_addr_match(const struct inet_bind2_bucket *tb2,void inet_bind_hash (struct sock *sk,struct inet_bind_bucket *tb,struct inet_bind2_bucket *tb2,unsigned short port) { -inet_sk (sk)->inet_num = port; +WRITE_ONCE (inet_sk (sk)->inet_num, port);inet_csk (sk)->icsk_bind_hash = tb;inet_csk (sk)->icsk_bind2_hash = tb2;sk_add_bind_node (sk, &tb2->owners);
@@ -224,7 +224,7 @@ static void __inet_put_port(struct sock *sk)spin_lock (&head->lock); tb =inet_csk (sk)->icsk_bind_hash;inet_csk (sk)->icsk_bind_hash =NULL ; -inet_sk (sk)->inet_num =0 ; +WRITE_ONCE (inet_sk (sk)->inet_num,0 ); sk->sk_userlocks &= ~SOCK_CONNECT_BIND ;spin_lock (&head2->lock);
@@ -352,7 +352,7 @@ static inline int compute_score(struct sock *sk, const struct net *net, {int score =-1 ; -if (net_eq (sock_net (sk), net) && sk->sk_num == hnum && +if (net_eq (sock_net (sk), net) &&READ_ONCE (sk->sk_num) == hnum && !ipv6_only_sock (sk)) {if (sk->sk_rcv_saddr != daddr)return -1 ;
@@ -1206,7 +1206,7 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row, sk->sk_hash =0 ;inet_sk (sk)->inet_sport =0 ; -inet_sk (sk)->inet_num =0 ; +WRITE_ONCE (inet_sk (sk)->inet_num,0 );if (tw)inet_twsk_bind_unhash (tw, hinfo);
net/ipv4/tcp_diag.c +1 -1
@@ -509,7 +509,7 @@ static void tcp_diag_dump(struct sk_buff *skb, struct netlink_callback *cb,if (r->sdiag_family !=AF_UNSPEC && sk->sk_family != r->sdiag_family) goto next_normal; -if (r->id.idiag_sport !=htons (sk->sk_num) && +if (r->id.idiag_sport !=htons (READ_ONCE (sk->sk_num)) && r->id.idiag_sport) goto next_normal; if (r->id.idiag_dport != sk->sk_dport &&
net/ipv6/inet6_hashtables.c +2 -1
@@ -95,7 +95,8 @@ static inline int compute_score(struct sock *sk, const struct net *net, {int score =-1 ; -if (net_eq (sock_net (sk), net) &&inet_sk (sk)->inet_num == hnum && +if (net_eq (sock_net (sk), net) && +READ_ONCE (inet_sk (sk)->inet_num) == hnum && sk->sk_family ==PF_INET6 ) {if (!ipv6_addr_equal (&sk->sk_v6_rcv_saddr, daddr))return -1 ;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 207/311 crypto: ccp - Fix use-after-free on error path
In reply to Sasha LevinFrom: Alper Ak <[email protected]>
[ Upstream commit 889b0e2721e793eb46cf7d17b965aa3252af3ec8 ]
In the error path of sev_tsm_init_locked(), the code dereferences 't' after it has been freed with kfree(). The pr_err() statement attempts to access t->tio_en and t->tio_init_done after the memory has been released.
Move the pr_err() call before kfree(t) to access the fields while the memory is still valid.
This issue reported by Smatch static analyser
--- drivers/crypto/ccp/sev-dev-tsm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/crypto/ccp/sev-dev-tsm.c +1 -1
@@ -378,9 +378,9 @@ void sev_tsm_init_locked(struct sev_device *sev, void *tio_status_page)return ; error_exit: -kfree (t);pr_err ("Failed to enable SEV-TIO: ret=%d en=%d initdone=%d SEV=%d\n" , ret, t->tio_en, t->tio_init_done,boot_cpu_has (X86_FEATURE_SEV )); +kfree (t); }void sev_tsm_uninit (struct sev_device *sev)
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 208/311 accel/amdxdna: Fill invalid payload for failed command
In reply to Sasha LevinFrom: Lizhi Hou <[email protected]>
[ Upstream commit 89ff45359abbf9d8d3c4aa3f5a57ed0be82b5a12 ]
Newer userspace applications may read the payload of a failed command to obtain detailed error information. However, the driver and old firmware versions may not support returning advanced error information. In this case, initialize the command payload with an invalid value so userspace can detect that no detailed error information is available.
--- drivers/accel/amdxdna/aie2_ctx.c | 23 ++++++++--------------- drivers/accel/amdxdna/amdxdna_ctx.c | 27 +++++++++++++++++++++++++++ drivers/accel/amdxdna/amdxdna_ctx.h | 3 +++ 3 files changed, 38 insertions(+), 15 deletions(-)
drivers/accel/amdxdna/aie2_ctx.c +8 -15
@@ -186,13 +186,13 @@ aie2_sched_resp_handler(void *handle, void __iomem *data, size_t size) cmd_abo = job->cmd_bo;if (unlikely (job->job_timeout)) { -amdxdna_cmd_set_state (cmd_abo,ERT_CMD_STATE_TIMEOUT ); +amdxdna_cmd_set_error (cmd_abo, job,0 ,ERT_CMD_STATE_TIMEOUT ); ret = -EINVAL ; goto out; }if (unlikely (!data) ||unlikely (size !=sizeof (u32))) { -amdxdna_cmd_set_state (cmd_abo,ERT_CMD_STATE_ABORT ); +amdxdna_cmd_set_error (cmd_abo, job,0 ,ERT_CMD_STATE_ABORT ); ret = -EINVAL ; goto out; }
@@ -202,7 +202,7 @@ aie2_sched_resp_handler(void *handle, void __iomem *data, size_t size)if (status ==AIE2_STATUS_SUCCESS )amdxdna_cmd_set_state (cmd_abo,ERT_CMD_STATE_COMPLETED );else -amdxdna_cmd_set_state (cmd_abo,ERT_CMD_STATE_ERROR ); +amdxdna_cmd_set_error (cmd_abo, job,0 ,ERT_CMD_STATE_ERROR ); out:aie2_sched_notify (job);
@@ -244,13 +244,13 @@ aie2_sched_cmdlist_resp_handler(void *handle, void __iomem *data, size_t size) cmd_abo = job->cmd_bo;if (unlikely (job->job_timeout)) { -amdxdna_cmd_set_state (cmd_abo,ERT_CMD_STATE_TIMEOUT ); +amdxdna_cmd_set_error (cmd_abo, job,0 ,ERT_CMD_STATE_TIMEOUT ); ret = -EINVAL ; goto out; }if (unlikely (!data) ||unlikely (size !=sizeof (u32) *3 )) { -amdxdna_cmd_set_state (cmd_abo,ERT_CMD_STATE_ABORT ); +amdxdna_cmd_set_error (cmd_abo, job,0 ,ERT_CMD_STATE_ABORT ); ret = -EINVAL ; goto out; }
@@ -270,19 +270,12 @@ aie2_sched_cmdlist_resp_handler(void *handle, void __iomem *data, size_t size) fail_cmd_idx, fail_cmd_status);if (fail_cmd_status ==AIE2_STATUS_SUCCESS ) { -amdxdna_cmd_set_state (cmd_abo,ERT_CMD_STATE_ABORT ); +amdxdna_cmd_set_error (cmd_abo, job, fail_cmd_idx,ERT_CMD_STATE_ABORT ); ret = -EINVAL ; - goto out; + }else { +amdxdna_cmd_set_error (cmd_abo, job, fail_cmd_idx,ERT_CMD_STATE_ERROR ); } -amdxdna_cmd_set_state (cmd_abo,ERT_CMD_STATE_ERROR ); -if (amdxdna_cmd_get_op (cmd_abo) ==ERT_CMD_CHAIN ) { -struct amdxdna_cmd_chain *cc =amdxdna_cmd_get_payload (cmd_abo,NULL ); - - cc->error_index = fail_cmd_idx; -if (cc->error_index >= cc->command_count) - cc->error_index =0 ; - } out:aie2_sched_notify (job);return ret;
drivers/accel/amdxdna/amdxdna_ctx.c +27 -0
@@ -135,6 +135,33 @@ u32 amdxdna_cmd_get_cu_idx(struct amdxdna_gem_obj *abo)return INVALID_CU_IDX ; } +int amdxdna_cmd_set_error (struct amdxdna_gem_obj *abo, +struct amdxdna_sched_job *job,u32 cmd_idx, +enum ert_cmd_state error_state) +{ +struct amdxdna_client *client = job->hwctx->client; +struct amdxdna_cmd *cmd = abo->mem.kva; +struct amdxdna_cmd_chain *cc =NULL ; + + cmd->header &= ~AMDXDNA_CMD_STATE ; + cmd->header |=FIELD_PREP (AMDXDNA_CMD_STATE , error_state); + +if (amdxdna_cmd_get_op (abo) ==ERT_CMD_CHAIN ) { + cc =amdxdna_cmd_get_payload (abo,NULL ); + cc->error_index = (cmd_idx < cc->command_count) ? cmd_idx :0 ; + abo =amdxdna_gem_get_obj (client, cc->data[0 ],AMDXDNA_BO_CMD ); +if (!abo) +return -EINVAL ; + cmd = abo->mem.kva; + } + +memset (cmd->data,0xff , abo->mem.size -sizeof (*cmd)); +if (cc) +amdxdna_gem_put_obj (abo); + +return 0 ; +} +/* * This should be called in close() and remove(). DO NOT call in other syscalls. * This guarantee that when hwctx and resources will be released, if user
drivers/accel/amdxdna/amdxdna_ctx.h +3 -0
@@ -167,6 +167,9 @@ amdxdna_cmd_get_state(struct amdxdna_gem_obj *abo)void *amdxdna_cmd_get_payload (struct amdxdna_gem_obj *abo,u32 *size);u32 amdxdna_cmd_get_cu_idx (struct amdxdna_gem_obj *abo); +int amdxdna_cmd_set_error (struct amdxdna_gem_obj *abo, +struct amdxdna_sched_job *job,u32 cmd_idx, +enum ert_cmd_state error_state);void amdxdna_sched_job_cleanup (struct amdxdna_sched_job *job);void amdxdna_hwctx_remove_all (struct amdxdna_client *client);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 209/311 udp: Unhash auto-bound connected sk from 4-tuple hash table when disconnected.
In reply to Sasha LevinFrom: Kuniyuki Iwashima <[email protected]>
[ Upstream commit 6996a2d2d0a64808c19c98002aeb5d9d1b2df6a4 ]
Let's say we bind() an UDP socket to the wildcard address with a non-zero port, connect() it to an address, and disconnect it from the address.
bind() sets SOCK_BINDPORT_LOCK on sk->sk_userlocks (but not SOCK_BINDADDR_LOCK), and connect() calls udp_lib_hash4() to put the socket into the 4-tuple hash table.
Then, __udp_disconnect() calls sk->sk_prot->rehash(sk).
It computes a new hash based on the wildcard address and moves the socket to a new slot in the 4-tuple hash table, leaving a garbage in the chain that no packet hits.
Let's remove such a socket from 4-tuple hash table when disconnected.
Note that udp_sk(sk)->udp_portaddr_hash needs to be udpated after udp_hash4_dec(hslot2) in udp_unhash4().
--- net/ipv4/udp.c | 25 +++++++++++++++---------- 1 file changed, 15 insertions(+), 10 deletions(-)
net/ipv4/udp.c +15 -10
@@ -2268,7 +2268,6 @@ void udp_lib_rehash(struct sock *sk, u16 newhash, u16 newhash4) udp_sk(sk)->udp_port_hash); hslot2 =udp_hashslot2 (udptable,udp_sk (sk)->udp_portaddr_hash); nhslot2 =udp_hashslot2 (udptable, newhash); -udp_sk (sk)->udp_portaddr_hash = newhash;if (hslot2 != nhslot2 ||rcu_access_pointer (sk->sk_reuseport_cb)) {
@@ -2302,19 +2301,25 @@ void udp_lib_rehash(struct sock *sk, u16 newhash, u16 newhash4)if (udp_hashed4 (sk)) {spin_lock_bh (&hslot->lock); -udp_rehash4 (udptable, sk, newhash4); -if (hslot2 != nhslot2) { -spin_lock (&hslot2->lock); -udp_hash4_dec (hslot2); -spin_unlock (&hslot2->lock); - -spin_lock (&nhslot2->lock); -udp_hash4_inc (nhslot2); -spin_unlock (&nhslot2->lock); +if (inet_rcv_saddr_any (sk)) { +udp_unhash4 (udptable, sk); + }else { +udp_rehash4 (udptable, sk, newhash4); +if (hslot2 != nhslot2) { +spin_lock (&hslot2->lock); +udp_hash4_dec (hslot2); +spin_unlock (&hslot2->lock); + +spin_lock (&nhslot2->lock); +udp_hash4_inc (nhslot2); +spin_unlock (&nhslot2->lock); + } }spin_unlock_bh (&hslot->lock); } + +udp_sk (sk)->udp_portaddr_hash = newhash; } }EXPORT_IPV6_MOD (udp_lib_rehash);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 210/311 tcp: give up on stronger sk_rcvbuf checks (for now)
In reply to Sasha LevinFrom: Jakub Kicinski <[email protected]>
[ Upstream commit 026dfef287c07f37d4d4eef7a0b5a4bfdb29b32d ]
We hit another corner case which leads to TcpExtTCPRcvQDrop
Connections which send RPCs in the 20-80kB range over loopback experience spurious drops. The exact conditions for most of the drops I investigated are that:
- socket exchanged >1MB of data so its not completely fresh
- rcvbuf is around 128kB (default, hasn't grown)
- there is ~60kB of data in rcvq
- skb > 64kB arrives
The sum of skb->len (!) of both of the skbs (the one already in rcvq and the arriving one) is larger than rwnd.
My suspicion is that this happens because __tcp_select_window() rounds the rwnd up to (1 << wscale) if less than half of the rwnd has been consumed.
Eric suggests that given the number of Fixes we already have pointing to 1d2fbaad7cd8 it's probably time to give up on it, until a bigger revamp of rmem management.
Also while we could risk tweaking the rwnd math, there are other drops on workloads I investigated, after the commit in question, not explained by this phenomenon.
--- net/ipv4/tcp_input.c | 16 +--------------- 1 file changed, 1 insertion(+), 15 deletions(-)
net/ipv4/tcp_input.c +1 -15
@@ -5116,25 +5116,11 @@ static void tcp_ofo_queue(struct sock *sk)static bool tcp_prune_ofo_queue (struct sock *sk,const struct sk_buff *in_skb);static int tcp_prune_queue (struct sock *sk,const struct sk_buff *in_skb); -/* Check if this incoming skb can be added to socket receive queues -* while satisfying sk->sk_rcvbuf limit. -* -* In theory we should use skb->truesize, but this can cause problems -* when applications use too small SO_RCVBUF values. -* When LRO / hw gro is used, the socket might have a high tp->scaling_ratio, -* allowing RWIN to be close to available space. -* Whenever the receive queue gets full, we can receive a small packet -* filling RWIN, but with a high skb->truesize, because most NIC use 4K page -* plus sk_buff metadata even when receiving less than 1500 bytes of payload. -* -* Note that we use skb->len to decide to accept or drop this packet, -* but sk->sk_rmem_alloc is the sum of all skb->truesize. -*/ static bool tcp_can_ingest (const struct sock *sk,const struct sk_buff *skb) {unsigned int rmem =atomic_read (&sk->sk_rmem_alloc); -return rmem + skb->len <= sk->sk_rcvbuf; +return rmem <= sk->sk_rcvbuf; } static int tcp_try_rmem_schedule(struct sock *sk, const struct sk_buff *skb,
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 211/311 xsk: Fix fragment node deletion to prevent buffer leak
In reply to Sasha LevinFrom: "Nikhil P. Rao" <[email protected]>
[ Upstream commit 60abb0ac11dccd6b98fd9182bc5f85b621688861 ]
After commit b692bf9a7543 ("xsk: Get rid of xdp_buff_xsk::xskb_list_node"), the list_node field is reused for both the xskb pool list and the buffer free list, this causes a buffer leak as described below.
xp_free() checks if a buffer is already on the free list using list_empty(&xskb->list_node). When list_del() is used to remove a node from the xskb pool list, it doesn't reinitialize the node pointers. This means list_empty() will return false even after the node has been removed, causing xp_free() to incorrectly skip adding the buffer to the free list.
Fix this by using list_del_init() instead of list_del() in all fragment handling paths, this ensures the list node is reinitialized after removal, allowing the list_empty() to work correctly.
--- include/net/xdp_sock_drv.h | 6 +++--- net/xdp/xsk.c | 2 +- 2 files changed, 4 insertions(+), 4 deletions(-)
include/net/xdp_sock_drv.h +3 -3
@@ -122,7 +122,7 @@ static inline void xsk_buff_free(struct xdp_buff *xdp) goto out;list_for_each_entry_safe (pos, tmp, xskb_list, list_node) { -list_del (&pos->list_node); +list_del_init (&pos->list_node);xp_free (pos); }
@@ -157,7 +157,7 @@ static inline struct xdp_buff *xsk_buff_get_frag(const struct xdp_buff *first) frag =list_first_entry_or_null (&xskb->pool->xskb_list, struct xdp_buff_xsk, list_node);if (frag) { -list_del (&frag->list_node); +list_del_init (&frag->list_node); ret = &frag->xdp; }
@@ -168,7 +168,7 @@ static inline void xsk_buff_del_frag(struct xdp_buff *xdp) { struct xdp_buff_xsk *xskb = container_of(xdp, struct xdp_buff_xsk, xdp); -list_del (&xskb->list_node); +list_del_init (&xskb->list_node); }static inline struct xdp_buff *xsk_buff_get_head (struct xdp_buff *first)
net/xdp/xsk.c +1 -1
@@ -186,7 +186,7 @@ static int xsk_rcv_zc(struct xdp_sock *xs, struct xdp_buff *xdp, u32 len) err =__xsk_rcv_zc (xs, pos, len, contd);if (err) goto err; -list_del (&pos->list_node); +list_del_init (&pos->list_node); }return 0 ;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 212/311 xsk: Fix zero-copy AF_XDP fragment drop
In reply to Sasha LevinFrom: "Nikhil P. Rao" <[email protected]>
[ Upstream commit f7387d6579d65efd490a864254101cb665f2e7a7 ]
AF_XDP should ensure that only a complete packet is sent to application. In the zero-copy case, if the Rx queue gets full as fragments are being enqueued, the remaining fragments are dropped.
For the multi-buffer case, add a check to ensure that the Rx queue has enough space for all fragments of a packet before starting to enqueue them.
--- net/xdp/xsk.c | 24 +++++++++++++++--------- 1 file changed, 15 insertions(+), 9 deletions(-)
net/xdp/xsk.c +15 -9
@@ -167,25 +167,31 @@ static int xsk_rcv_zc(struct xdp_sock *xs, struct xdp_buff *xdp, u32 len)struct xdp_buff_xsk *pos, *tmp;struct list_head *xskb_list;u32 contd =0 ; +u32 num_desc;int err; -if (frags) - contd =XDP_PKT_CONTD ; +if (likely (!frags)) { + err =__xsk_rcv_zc (xs, xskb, len, contd); +if (err) + goto err; +return 0 ; + } - err =__xsk_rcv_zc (xs, xskb, len, contd); -if (err) + contd =XDP_PKT_CONTD ; + num_desc =xdp_get_shared_info_from_buff (xdp)->nr_frags +1 ; +if (xskq_prod_nb_free (xs->rx, num_desc) < num_desc) { + xs->rx_queue_full++; + err = -ENOBUFS ; goto err; -if (likely (!frags)) -return 0 ; + } +__xsk_rcv_zc (xs, xskb, len, contd); xskb_list = &xskb->pool->xskb_list;list_for_each_entry_safe (pos, tmp, xskb_list, list_node) {if (list_is_singular (xskb_list)) contd =0 ; len = pos->xdp.data_end - pos->xdp.data; - err =__xsk_rcv_zc (xs, pos, len, contd); -if (err) - goto err; +__xsk_rcv_zc (xs, pos, len, contd);list_del_init (&pos->list_node); }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 213/311 dpaa2-switch: Fix interrupt storm after receiving bad if_id in IRQ handler
In reply to Sasha LevinFrom: Guenter Roeck <[email protected]>
[ Upstream commit 74badb9c20b1a9c02a95c735c6d3cd6121679c93 ]
Commit 31a7a0bbeb00 ("dpaa2-switch: add bounds check for if_id in IRQ handler") introduces a range check for if_id to avoid an out-of-bounds access. If an out-of-bounds if_id is detected, the interrupt status is not cleared. This may result in an interrupt storm.
Clear the interrupt status after detecting an out-of-bounds if_id to avoid the problem.
Found by an experimental AI code review agent at Google.
--- drivers/net/ethernet/freescale/dpaa2/dpaa2-switch.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/net/ethernet/freescale/dpaa2/dpaa2-switch.c +2 -1
@@ -1533,7 +1533,7 @@ static irqreturn_t dpaa2_switch_irq0_handler_thread(int irq_num, void *arg) if_id = (status &0xFFFF0000 ) >>16 ;if (if_id >= ethsw->sw_attr.num_ifs) {dev_err (dev,"Invalid if_id %d in IRQ status\n" , if_id); - goto out; + goto out_clear; } port_priv = ethsw->ports[if_id];
@@ -1553,6 +1553,7 @@ static irqreturn_t dpaa2_switch_irq0_handler_thread(int irq_num, void *arg)dpaa2_switch_port_connect_mac (port_priv); } +out_clear: err =dpsw_clear_irq_status (ethsw->mc_io,0 , ethsw->dpsw_handle,DPSW_IRQ_INDEX_IF , status);if (err)
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 214/311 atm: lec: fix null-ptr-deref in lec_arp_clear_vccs
In reply to Sasha LevinFrom: Jiayuan Chen <[email protected]>
[ Upstream commit 101bacb303e89dc2e0640ae6a5e0fb97c4eb45bb ]
syzkaller reported a null-ptr-deref in lec_arp_clear_vccs(). This issue can be easily reproduced using the syzkaller reproducer.
In the ATM LANE (LAN Emulation) module, the same atm_vcc can be shared by multiple lec_arp_table entries (e.g., via entry->vcc or entry->recv_vcc). When the underlying VCC is closed, lec_vcc_close() iterates over all ARP entries and calls lec_arp_clear_vccs() for each matched entry.
For example, when lec_vcc_close() iterates through the hlists in priv->lec_arp_empty_ones or other ARP tables:
1. In the first iteration, for the first matched ARP entry sharing the VCC, lec_arp_clear_vccs() frees the associated vpriv (which is vcc->user_back) and sets vcc->user_back to NULL.
2. In the second iteration, for the next matched ARP entry sharing the same VCC, lec_arp_clear_vccs() is called again. It obtains a NULL vpriv from vcc->user_back (via LEC_VCC_PRIV(vcc)) and then attempts to dereference it via vcc->pop = vpriv->old_pop, leading to a null-ptr-deref crash.
Fix this by adding a null check for vpriv before dereferencing it. If vpriv is already NULL, it means the VCC has been cleared by a previous call, so we can safely skip the cleanup and just clear the entry's vcc/recv_vcc pointers.
The entire cleanup block (including vcc_release_async()) is placed inside the vpriv guard because a NULL vpriv indicates the VCC has already been fully released by a prior iteration — repeating the teardown would redundantly set flags and trigger callbacks on an already-closing socket.
The Fixes tag points to the initial commit because the entry->vcc path has been vulnerable since the original code. The entry->recv_vcc path was later added by commit 8d9f73c0ad2f ("atm: fix a memory leak of vcc->user_back") with the same pattern, and both paths are fixed here.
--- net/atm/lec.c | 26 +++++++++++++++----------- 1 file changed, 15 insertions(+), 11 deletions(-)
net/atm/lec.c +15 -11
@@ -1260,24 +1260,28 @@ static void lec_arp_clear_vccs(struct lec_arp_table *entry)struct lec_vcc_priv *vpriv =LEC_VCC_PRIV (vcc);struct net_device *dev = (struct net_device *)vcc->proto_data; - vcc->pop = vpriv->old_pop; -if (vpriv->xoff) -netif_wake_queue (dev); -kfree (vpriv); - vcc->user_back =NULL ; - vcc->push = entry->old_push; -vcc_release_async (vcc, -EPIPE ); +if (vpriv) { + vcc->pop = vpriv->old_pop; +if (vpriv->xoff) +netif_wake_queue (dev); +kfree (vpriv); + vcc->user_back =NULL ; + vcc->push = entry->old_push; +vcc_release_async (vcc, -EPIPE ); + } entry->vcc =NULL ; }if (entry->recv_vcc) {struct atm_vcc *vcc = entry->recv_vcc;struct lec_vcc_priv *vpriv =LEC_VCC_PRIV (vcc); -kfree (vpriv); - vcc->user_back =NULL ; +if (vpriv) { +kfree (vpriv); + vcc->user_back =NULL ; - entry->recv_vcc->push = entry->old_recv_push; -vcc_release_async (entry->recv_vcc, -EPIPE ); + entry->recv_vcc->push = entry->old_recv_push; +vcc_release_async (entry->recv_vcc, -EPIPE ); + } entry->recv_vcc =NULL ; } }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 215/311 net: ti: icssg-prueth: Fix ping failure after offload mode setup when link speed is not 1G
In reply to Sasha LevinFrom: MD Danish Anwar <[email protected]>
[ Upstream commit 147792c395db870756a0dc87ce656c75ae7ab7e8 ]
When both eth interfaces with links up are added to a bridge or hsr interface, ping fails if the link speed is not 1Gbps (e.g., 100Mbps).
The issue is seen because when switching to offload (bridge/hsr) mode, prueth_emac_restart() restarts the firmware and clears DRAM with memset_io(), setting all memory to 0. This includes PORT_LINK_SPEED_OFFSET which firmware reads for link speed. The value 0 corresponds to FW_LINK_SPEED_1G (0x00), so for 1Gbps links the default value is correct and ping works. For 100Mbps links, the firmware needs FW_LINK_SPEED_100M (0x01) but gets 0 instead, causing ping to fail. The function emac_adjust_link() is called to reconfigure, but it detects no state change (emac->link is still 1, speed/duplex match PHY) so new_state remains false and icssg_config_set_speed() is never called to correct the firmware speed value.
The fix resets emac->link to 0 before calling emac_adjust_link() in prueth_emac_common_start(). This forces new_state=true, ensuring icssg_config_set_speed() is called to write the correct speed value to firmware memory.
--- drivers/net/ethernet/ti/icssg/icssg_prueth.c | 8 ++++++++ 1 file changed, 8 insertions(+)
drivers/net/ethernet/ti/icssg/icssg_prueth.c +8 -0
@@ -273,6 +273,14 @@ static int prueth_emac_common_start(struct prueth *prueth)if (ret) goto disable_class; +/* Reset link state to force reconfiguration in +* emac_adjust_link(). Without this, if the link was already up +* before restart, emac_adjust_link() won't detect any state +* change and will skip critical configuration like writing +* speed to firmware. +*/ + emac->link =0 ; +mutex_lock (&emac->ndev->phydev->lock);emac_adjust_link (emac->ndev);mutex_unlock (&emac->ndev->phydev->lock);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 216/311 amd-xgbe: fix MAC_TCR_SS register width for 2.5G and 10M speeds
In reply to Sasha LevinFrom: Raju Rangoju <[email protected]>
[ Upstream commit 9439a661c2e80485406ce2c90b107ca17858382d ]
Extend the MAC_TCR_SS (Speed Select) register field width from 2 bits to 3 bits to properly support all speed settings.
The MAC_TCR register's SS field encoding requires 3 bits to represent all supported speeds:
- 0x00: 10Gbps (XGMII)
- 0x02: 2.5Gbps (GMII) / 100Mbps
- 0x03: 1Gbps / 10Mbps
- 0x06: 2.5Gbps (XGMII) - P100a only
With only 2 bits, values 0x04-0x07 cannot be represented, which breaks 2.5G XGMII mode on newer platforms and causes incorrect speed select values to be programmed.
--- drivers/net/ethernet/amd/xgbe/xgbe-common.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/amd/xgbe/xgbe-common.h +1 -1
@@ -431,7 +431,7 @@#define MAC_SSIR_SSINC_INDEX 16#define MAC_SSIR_SSINC_WIDTH 8#define MAC_TCR_SS_INDEX 29 -#define MAC_TCR_SS_WIDTH 2 +#define MAC_TCR_SS_WIDTH 3#define MAC_TCR_TE_INDEX 0#define MAC_TCR_TE_WIDTH 1#define MAC_TCR_VNE_INDEX 24
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 217/311 regulator: mt6363: Fix incorrect and redundant IRQ disposal in probe
In reply to Sasha LevinFrom: Felix Gu <[email protected]>
[ Upstream commit 23942b71f07cc99e39d9216a5b370df494759d8c ]
In mt6363_regulator_probe(), devm_add_action_or_reset() is used to automatically dispose of the IRQ mapping if the probe fails or the device is removed.
The manual call to irq_dispose_mapping() in the error path was redundant as the reset action already triggers mt6363_irq_remove(). Furthermore, the manual call incorrectly passed the hardware IRQ number (info->hwirq) instead of the virtual IRQ mapping (info->virq).
Remove the redundant and incorrect manual disposal.
--- drivers/regulator/mt6363-regulator.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-)
drivers/regulator/mt6363-regulator.c +1 -3
@@ -899,10 +899,8 @@ static int mt6363_regulator_probe(struct platform_device *pdev) "Failed to map IRQ%d\n", info->hwirq); ret =devm_add_action_or_reset (dev, mt6363_irq_remove, &info->virq); -if (ret) { -irq_dispose_mapping (info->hwirq); +if (ret)return ret; - } config.driver_data = info;INIT_DELAYED_WORK (&info->oc_work, mt6363_oc_irq_enable_work);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 218/311 can: bcm: fix locking for bcm_op runtime updates
In reply to Sasha LevinFrom: Oliver Hartkopp <[email protected]>
[ Upstream commit c35636e91e392e1540949bbc67932167cb48bc3a ]
Commit c2aba69d0c36 ("can: bcm: add locking for bcm_op runtime updates") added a locking for some variables that can be modified at runtime when updating the sending bcm_op with a new TX_SETUP command in bcm_tx_setup().
Usually the RX_SETUP only handles and filters incoming traffic with one
exception: When the RX_RTR_FRAME flag is set a predefined CAN frame is sent when a specific RTR frame is received. Therefore the rx bcm_op uses bcm_can_tx() which uses the bcm_tx_lock that was only initialized in bcm_tx_setup(). Add the missing spin_lock_init() when allocating the bcm_op in bcm_rx_setup() to handle the RTR case properly.
--- net/can/bcm.c | 1 + 1 file changed, 1 insertion(+)
net/can/bcm.c +1 -0
@@ -1170,6 +1170,7 @@ static int bcm_rx_setup(struct bcm_msg_head *msg_head, struct msghdr *msg,if (!op)return -ENOMEM ; +spin_lock_init (&op->bcm_tx_lock); op->can_id = msg_head->can_id; op->nframes = msg_head->nframes; op->cfsiz =CFSIZ (msg_head->flags);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 219/311 can: dummy_can: dummy_can_init(): fix packet statistics
In reply to Sasha LevinFrom: Oliver Hartkopp <[email protected]>
[ Upstream commit c77bfbdd6aac31b152ee81522cd90ad1de18738f ]
The former implementation was only counting the tx_packets value but not the tx_bytes as the skb was dropped on driver layer.
Enable CAN echo support (IFF_ECHO) in dummy_can_init(), which activates the code for setting and retrieving the echo SKB and counts the tx_bytes correctly.
--- drivers/net/can/dummy_can.c | 1 + 1 file changed, 1 insertion(+)
drivers/net/can/dummy_can.c +1 -0
@@ -241,6 +241,7 @@ static int __init dummy_can_init(void) dev->netdev_ops = &dummy_can_netdev_ops; dev->ethtool_ops = &dummy_can_ethtool_ops; + dev->flags |=IFF_ECHO ;/* enable echo handling */ priv =netdev_priv (dev); priv->can.bittiming_const = &dummy_can_bittiming_const; priv->can.bitrate_max =20 *MEGA /* BPS */ ;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 220/311 can: mcp251x: fix deadlock in error path of mcp251x_open
In reply to Sasha LevinFrom: Alban Bedel <[email protected]>
[ Upstream commit ab3f894de216f4a62adc3b57e9191888cbf26885 ]
The mcp251x_open() function call free_irq() in its error path with the mpc_lock mutex held. But if an interrupt already occurred the interrupt handler will be waiting for the mpc_lock and free_irq() will deadlock waiting for the handler to finish.
This issue is similar to the one fixed in commit 7dd9c26bd6cf ("can:
mcp251x: fix deadlock if an interrupt occurs during mcp251x_open") but for the error path.
To solve this issue move the call to free_irq() after the lock is released. Setting priv->force_quit = 1 beforehand ensure that the IRQ handler will exit right away once it acquired the lock.
--- drivers/net/can/spi/mcp251x.c | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-)
drivers/net/can/spi/mcp251x.c +14 -1
@@ -1214,6 +1214,7 @@ static int mcp251x_open(struct net_device *net) {struct mcp251x_priv *priv =netdev_priv (net);struct spi_device *spi = priv->spi; +bool release_irq = false;unsigned long flags =0 ;int ret;
@@ -1257,12 +1258,24 @@ static int mcp251x_open(struct net_device *net)return 0 ; out_free_irq: -free_irq (spi->irq, priv); +/* The IRQ handler might be running, and if so it will be waiting +* for the lock. But free_irq() must wait for the handler to finish +* so calling it here would deadlock. +* +* Setting priv->force_quit will let the handler exit right away +* without any access to the hardware. This make it safe to call +* free_irq() after the lock is released. +*/ + priv->force_quit =1 ; + release_irq = true; +mcp251x_hw_sleep (spi); out_close:mcp251x_power_enable (priv->transceiver,0 );close_candev (net);mutex_unlock (&priv->mcp_lock); +if (release_irq) +free_irq (spi->irq, priv);return ret; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 221/311 wifi: rsi: Don't default to -EOPNOTSUPP in rsi_mac80211_config
In reply to Sasha LevinFrom: Sebastian Krzyszkowiak <[email protected]>
[ Upstream commit d973b1039ccde6b241b438d53297edce4de45b5c ]
This triggers a WARN_ON in ieee80211_hw_conf_init and isn't the expected behavior from the driver - other drivers default to 0 too.
--- drivers/net/wireless/rsi/rsi_91x_mac80211.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/wireless/rsi/rsi_91x_mac80211.c +1 -1
@@ -668,7 +668,7 @@ static int rsi_mac80211_config(struct ieee80211_hw *hw,struct rsi_hw *adapter = hw->priv;struct rsi_common *common = adapter->priv;struct ieee80211_conf *conf = &hw->conf; -int status = -EOPNOTSUPP ; +int status =0 ;mutex_lock (&common->mutex);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 222/311 drm/syncobj: Fix handle <-> fd ioctls with dirty stack
In reply to Sasha LevinFrom: Julian Orth <[email protected]>
[ Upstream commit 2e3649e237237258a08d75afef96648dd2b379f7 ]
Consider the following application:
#include <fcntl.h>
#include <string.h>
#include <drm/drm.h>
#include <sys/ioctl.h>
int main(void) {
int fd = open("/dev/dri/renderD128", O_RDWR);
struct drm_syncobj_create arg1;
ioctl(fd, DRM_IOCTL_SYNCOBJ_CREATE, &arg1);
struct drm_syncobj_handle arg2;
memset(&arg2, 1, sizeof(arg2)); // simulate dirty stack
arg2.handle = arg1.handle;
arg2.flags = 0;
arg2.fd = 0;
arg2.pad = 0;
// arg2.point = 0; // userspace is required to set point to 0
ioctl(fd, DRM_IOCTL_SYNCOBJ_HANDLE_TO_FD, &arg2);
}The last ioctl returns EINVAL because args->point is not 0. However, userspace developed against older kernel versions is not aware of the new point field and might therefore not initialize it.
The correct check would be
if (args->flags & DRM_SYNCOBJ_FD_TO_HANDLE_FLAGS_TIMELINE)
return -EINVAL;However, there might already be userspace that relies on this not returning an error as long as point == 0. Therefore use the more lenient check.
--- drivers/gpu/drm/drm_syncobj.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/drm_syncobj.c +2 -2
@@ -900,7 +900,7 @@ drm_syncobj_handle_to_fd_ioctl(struct drm_device *dev, void *data,return drm_syncobj_export_sync_file (file_private, args->handle, point, &args->fd); -if (args->point) +if (point)return -EINVAL ; return drm_syncobj_handle_to_fd(file_private, args->handle,
@@ -934,7 +934,7 @@ drm_syncobj_fd_to_handle_ioctl(struct drm_device *dev, void *data, args->handle, point); -if (args->point) +if (point)return -EINVAL ; return drm_syncobj_fd_to_handle(file_private, args->fd,
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 223/311 drm/xe: Do not preempt fence signaling CS instructions
In reply to Sasha LevinFrom: Matthew Brost <[email protected]>
[ Upstream commit cdc8a1e11f4d5b480ec750e28010c357185b95a6 ]
If a batch buffer is complete, it makes little sense to preempt the fence signaling instructions in the ring, as the largest portion of the work (the batch buffer) is already done and fence signaling consists of only a few instructions. If these instructions are preempted, the GuC would need to perform a context switch just to signal the fence, which is costly and delays fence signaling. Avoid this scenario by disabling preemption immediately after the BB start instruction and re-enabling it after executing the fence signaling instructions.
--- drivers/gpu/drm/xe/xe_ring_ops.c | 9 +++++++++ 1 file changed, 9 insertions(+)
drivers/gpu/drm/xe/xe_ring_ops.c +9 -0
@@ -267,6 +267,9 @@ static void __emit_job_gen12_simple(struct xe_sched_job *job, struct xe_lrc *lrc i =emit_bb_start (batch_addr, ppgtt_flag, dw, i); +/* Don't preempt fence signaling */ + dw[i++] =MI_ARB_ON_OFF |MI_ARB_DISABLE ; +if (job->user_fence.used) { i =emit_flush_dw (dw, i); i = emit_store_imm_ppgtt_posted(job->user_fence.addr,
@@ -332,6 +335,9 @@ static void __emit_job_gen12_video(struct xe_sched_job *job, struct xe_lrc *lrc, i =emit_bb_start (batch_addr, ppgtt_flag, dw, i); +/* Don't preempt fence signaling */ + dw[i++] =MI_ARB_ON_OFF |MI_ARB_DISABLE ; +if (job->user_fence.used) { i =emit_flush_dw (dw, i); i = emit_store_imm_ppgtt_posted(job->user_fence.addr,
@@ -384,6 +390,9 @@ static void __emit_job_gen12_render_compute(struct xe_sched_job *job, i =emit_bb_start (batch_addr, ppgtt_flag, dw, i); +/* Don't preempt fence signaling */ + dw[i++] =MI_ARB_ON_OFF |MI_ARB_DISABLE ; + i =emit_render_cache_flush (job, dw, i);if (job->user_fence.used)
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 224/311 drm/xe/configfs: Free ctx_restore_mid_bb in release
In reply to Sasha LevinFrom: Shuicheng Lin <[email protected]>
[ Upstream commit e377182f0266f46f02d01838e6bde67b9dac0d66 ]
ctx_restore_mid_bb memory is allocated in wa_bb_store(), but xe_config_device_release() only frees ctx_restore_post_bb.
Free ctx_restore_mid_bb[0].cs as well to avoid leaking the allocation when the configfs device is removed.
--- drivers/gpu/drm/xe/xe_configfs.c | 1 + 1 file changed, 1 insertion(+)
drivers/gpu/drm/xe/xe_configfs.c +1 -0
@@ -830,6 +830,7 @@ static void xe_config_device_release(struct config_item *item)mutex_destroy (&dev->lock); +kfree (dev->config.ctx_restore_mid_bb[0 ].cs);kfree (dev->config.ctx_restore_post_bb[0 ].cs);kfree (dev); }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 225/311 drm/xe/queue: Call fini on exec queue creation fail
In reply to Sasha LevinFrom: Tomasz Lis <[email protected]>
[ Upstream commit 99f9b5343cae80eb0dfe050baf6c86d722b3ba2e ]
Every call to queue init should have a corresponding fini call. Skipping this would mean skipping removal of the queue from GuC list (which is part of guc_id allocation). A damaged queue stored in exec_queue_lookup list would lead to invalid memory reference, sooner or later.
Call fini to free guc_id. This must be done before any internal LRCs are freed.
Since the finalization with this extra call became very similar to __xe_exec_queue_fini(), reuse that. To make this reuse possible, alter xe_lrc_put() so it can survive NULL parameters, like other similar functions.
v2: Reuse _xe_exec_queue_fini(). Make xe_lrc_put() aware of NULLs.
--- drivers/gpu/drm/xe/xe_exec_queue.c | 23 +++++++++++------------ drivers/gpu/drm/xe/xe_lrc.h | 3 ++- 2 files changed, 13 insertions(+), 13 deletions(-)
drivers/gpu/drm/xe/xe_exec_queue.c +11 -12
@@ -185,6 +185,16 @@ static struct xe_exec_queue *__xe_exec_queue_alloc(struct xe_device *xe,return q; } +static void __xe_exec_queue_fini (struct xe_exec_queue *q) +{ +int i; + + q->ops->fini (q); + +for (i =0 ; i < q->width; ++i) +xe_lrc_put (q->lrc[i]); +} +static int __xe_exec_queue_init (struct xe_exec_queue *q,u32 exec_queue_flags) {int i, err;
@@ -239,21 +249,10 @@ static int __xe_exec_queue_init(struct xe_exec_queue *q, u32 exec_queue_flags)return 0 ; err_lrc: -for (i = i -1 ; i >=0 ; --i) -xe_lrc_put (q->lrc[i]); +__xe_exec_queue_fini (q);return err; } -static void __xe_exec_queue_fini (struct xe_exec_queue *q) -{ -int i; - - q->ops->fini (q); - -for (i =0 ; i < q->width; ++i) -xe_lrc_put (q->lrc[i]); -} - struct xe_exec_queue *xe_exec_queue_create(struct xe_device *xe, struct xe_vm *vm, u32 logical_mask, u16 width, struct xe_hw_engine *hwe, u32 flags,
drivers/gpu/drm/xe/xe_lrc.h +2 -1
@@ -73,7 +73,8 @@ static inline struct xe_lrc *xe_lrc_get(struct xe_lrc *lrc)*/ static inline void xe_lrc_put (struct xe_lrc *lrc) { -kref_put (&lrc->refcount, xe_lrc_destroy); +if (lrc) +kref_put (&lrc->refcount, xe_lrc_destroy); }/**
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 226/311 blktrace: fix __this_cpu_read/write in preemptible context
In reply to Sasha LevinFrom: Chaitanya Kulkarni <[email protected]>
[ Upstream commit da46b5dfef48658d03347cda21532bcdbb521e67 ]
tracing_record_cmdline() internally uses __this_cpu_read() and __this_cpu_write() on the per-CPU variable trace_cmdline_save, and trace_save_cmdline() explicitly asserts preemption is disabled via lockdep_assert_preemption_disabled(). These operations are only safe when preemption is off, as they were designed to be called from the scheduler context (probe_wakeup_sched_switch() / probe_wakeup()).
__blk_add_trace() was calling tracing_record_cmdline(current) early in the blk_tracer path, before ring buffer reservation, from process context where preemption is fully enabled. This triggers the following using blktests/blktrace/002:
blktrace/002 (blktrace ftrace corruption with sysfs trace) [failed]
runtime 0.367s ... 0.437s
something found in dmesg:
[ 81.211018] run blktests blktrace/002 at 2026-02-25 22:24:33
[ 81.239580] null_blk: disk nullb1 created
[ 81.357294] BUG: using __this_cpu_read() in preemptible [00000000] code: dd/2516
[ 81.362842] caller is tracing_record_cmdline+0x10/0x40
[ 81.362872] CPU: 16 UID: 0 PID: 2516 Comm: dd Tainted: G N 7.0.0-rc1lblk+ #84 PREEMPT(full)
[ 81.362877] Tainted: [N]=TEST
[ 81.362878] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.17.0-0-gb52ca86e094d-prebuilt.qemu.org 04/01/2014
[ 81.362881] Call Trace:
[ 81.362884] <TASK>
[ 81.362886] dump_stack_lvl+0x8d/0xb0
...
(See '/mnt/sda/blktests/results/nodev/blktrace/002.dmesg' for the entire message)[ 81.211018] run blktests blktrace/002 at 2026-02-25 22:24:33 [ 81.239580] null_blk: disk nullb1 created [ 81.357294] BUG: using __this_cpu_read() in preemptible [00000000] code: dd/2516 [ 81.362842] caller is tracing_record_cmdline+0x10/0x40 [ 81.362872] CPU: 16 UID: 0 PID: 2516 Comm: dd Tainted: G N 7.0.0-rc1lblk+ #84 PREEMPT(full) [ 81.362877] Tainted: [N]=TEST [ 81.362878] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.17.0-0-gb52ca86e094d-prebuilt.qemu.org 04/01/2014 [ 81.362881] Call Trace: [ 81.362884] <TASK> [ 81.362886] dump_stack_lvl+0x8d/0xb0 [ 81.362895] check_preemption_disabled+0xce/0xe0 [ 81.362902] tracing_record_cmdline+0x10/0x40 [ 81.362923] __blk_add_trace+0x307/0x5d0 [ 81.362934] ? lock_acquire+0xe0/0x300 [ 81.362940] ? iov_iter_extract_pages+0x101/0xa30 [ 81.362959] blk_add_trace_bio+0x106/0x1e0 [ 81.362968] submit_bio_noacct_nocheck+0x24b/0x3a0 [ 81.362979] ? lockdep_init_map_type+0x58/0x260 [ 81.362988] submit_bio_wait+0x56/0x90 [ 81.363009] __blkdev_direct_IO_simple+0x16c/0x250 [ 81.363026] ? __pfx_submit_bio_wait_endio+0x10/0x10 [ 81.363038] ? rcu_read_lock_any_held+0x73/0xa0 [ 81.363051] blkdev_read_iter+0xc1/0x140 [ 81.363059] vfs_read+0x20b/0x330 [ 81.363083] ksys_read+0x67/0xe0 [ 81.363090] do_syscall_64+0xbf/0xf00 [ 81.363102] entry_SYSCALL_64_after_hwframe+0x76/0x7e [ 81.363106] RIP: 0033:0x7f281906029d [ 81.363111] Code: 31 c0 e9 c6 fe ff ff 50 48 8d 3d 66 63 0a 00 e8 59 ff 01 00 66 0f 1f 84 00 00 00 00 00 80 3d 41 33 0e 00 00 74 17 31 c0 0f 05 <48> 3d 00 f0 ff ff 77 5b c3 66 2e 0f 1f 84 00 00 00 00 00 48 83 ec [ 81.363113] RSP: 002b:00007ffca127dd48 EFLAGS: 00000246 ORIG_RAX: 0000000000000000 [ 81.363120] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f281906029d [ 81.363122] RDX: 0000000000001000 RSI: 0000559f8bfae000 RDI: 0000000000000000 [ 81.363123] RBP: 0000000000001000 R08: 0000002863a10a81 R09: 00007f281915f000 [ 81.363124] R10: 00007f2818f77b60 R11: 0000000000000246 R12: 0000559f8bfae000 [ 81.363126] R13: 0000000000000000 R14: 0000000000000000 R15: 000000000000000a [ 81.363142] </TASK>
The same BUG fires from blk_add_trace_plug(), blk_add_trace_unplug(), and blk_add_trace_rq() paths as well.
The purpose of tracing_record_cmdline() is to cache the task->comm for a given PID so that the trace can later resolve it. It is only meaningful when a trace event is actually being recorded. Ring buffer reservation via ring_buffer_lock_reserve() disables preemption, and preemption remains disabled until the event is committed :-
__blk_add_trace()
__trace_buffer_lock_reserve()
__trace_buffer_lock_reserve()
ring_buffer_lock_reserve()
preempt_disable_notrace(); <---With this fix blktests for blktrace pass:
blktests (master) # ./check blktrace
blktrace/001 (blktrace zone management command tracing) [passed]
runtime 3.650s ... 3.647s
blktrace/002 (blktrace ftrace corruption with sysfs trace) [passed]
runtime 0.411s ... 0.384s--- kernel/trace/blktrace.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
kernel/trace/blktrace.c +1 -2
@@ -383,8 +383,6 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes, cpu =raw_smp_processor_id ();if (blk_tracer) { -tracing_record_cmdline (current); - buffer = blk_tr->array_buffer.buffer; trace_ctx =tracing_gen_ctx_flags (0 );switch (bt->version) {
@@ -419,6 +417,7 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,if (!event)return ; +tracing_record_cmdline (current);switch (bt->version) {case 1 : record_blktrace_event(ring_buffer_event_data(event),
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 227/311 rust: kunit: fix warning when !CONFIG_PRINTK
In reply to Sasha LevinFrom: Alexandre Courbot <[email protected]>
[ Upstream commit 7dd34dfc8dfa92a7244242098110388367996ac3 ]
If CONFIG_PRINTK is not set, then the following warnings are issued during build:
warning: unused variable: args
--> ../rust/kernel/kunit.rs:16:12
| 16 | pub fn err(args: fmt::Arguments<'_>) {
| ^^^^ help: if this is intentional, prefix it with an underscore: _args
|
= note: #[warn(unused_variables)] (part of #[warn(unused)]) on by default
warning: unused variable: args
--> ../rust/kernel/kunit.rs:32:13
| 32 | pub fn info(args: fmt::Arguments<'_>) {
| ^^^^ help: if this is intentional, prefix it with an underscore: _args
Fix this by adding a no-op assignment using args when CONFIG_PRINTK is not set.
--- rust/kernel/kunit.rs | 8 ++++++++ 1 file changed, 8 insertions(+)
rust/kernel/kunit.rs +8 -0
@@ -17,6 +17,10 @@/// Public but hidden since it should only be used from KUnit generated code. #[doc(hidden)] pub fn err (args: fmt::Arguments <'_>) { +// `args` is unused if `CONFIG_PRINTK` is not set - this avoids a build-time warning. +#[cfg(not( CONFIG_PRINTK ))] +let _ = args; +// SAFETY: The format string is null-terminated and the `%pA` specifier matches the argument we // are passing. #[cfg( CONFIG_PRINTK )]
@@ -33,6 +37,10 @@ pub fn err(args: fmt::Arguments<'_>) {/// Public but hidden since it should only be used from KUnit generated code. #[doc(hidden)] pub fn info (args: fmt::Arguments <'_>) { +// `args` is unused if `CONFIG_PRINTK` is not set - this avoids a build-time warning. +#[cfg(not( CONFIG_PRINTK ))] +let _ = args; +// SAFETY: The format string is null-terminated and the `%pA` specifier matches the argument we // are passing. #[cfg( CONFIG_PRINTK )]
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 228/311 kunit: tool: copy caller args in run_kernel to prevent mutation
In reply to Sasha LevinFrom: Shuvam Pandey <[email protected]>
[ Upstream commit 40804c4974b8df2adab72f6475d343eaff72b7f6 ]
run_kernel() appended KUnit flags directly to the caller-provided args list. When exec_tests() calls run_kernel() repeatedly (e.g. with --run_isolated), each call mutated the same list, causing later runs to inherit stale filter_glob values and duplicate kunit.enable flags.
Fix this by copying args at the start of run_kernel(). Add a regression test that calls run_kernel() twice with the same list and verifies the original remains unchanged.
--- tools/testing/kunit/kunit_kernel.py | 6 ++++-- tools/testing/kunit/kunit_tool_test.py | 26 ++++++++++++++++++++++++++ 2 files changed, 30 insertions(+), 2 deletions(-)
tools/testing/kunit/kunit_kernel.py +4 -2
@@ -346,8 +346,10 @@ class LinuxSourceTree: return self.validate_config(build_dir) def run_kernel(self, args: Optional[List[str]]=None, build_dir: str='', filter_glob: str='', filter: str='', filter_action: Optional[str]=None, timeout: Optional[int]=None) -> Iterator[str]: - if not args: - args = [] + # Copy to avoid mutating the caller-supplied list. exec_tests() reuses + # the same args across repeated run_kernel() calls (e.g. --run_isolated), + # so appending to the original would accumulate stale flags on each call. + args = list(args) if args else [] if filter_glob: args.append('kunit.filter_glob=' + filter_glob) if filter:
tools/testing/kunit/kunit_tool_test.py +26 -0
@@ -489,6 +489,32 @@ class LinuxSourceTreeTest(unittest.TestCase): with open(kunit_kernel.get_outfile_path(build_dir), 'rt') as outfile: self.assertEqual(outfile.read(), 'hi\nbye\n', msg='Missing some output') + def test_run_kernel_args_not_mutated(self): + """Verify run_kernel() copies args so callers can reuse them.""" + start_calls = [] + + def fake_start(start_args, unused_build_dir): + start_calls.append(list(start_args)) + return subprocess.Popen(['printf', 'KTAP version 1\n'], + text=True, stdout=subprocess.PIPE) + + with tempfile.TemporaryDirectory('') as build_dir: + tree = kunit_kernel.LinuxSourceTree(build_dir, + kunitconfig_paths=[os.devnull]) + with mock.patch.object(tree._ops, 'start', side_effect=fake_start), \ + mock.patch.object(kunit_kernel.subprocess, 'call'): + kernel_args = ['mem=1G'] + for _ in tree.run_kernel(args=kernel_args, build_dir=build_dir, + filter_glob='suite.test1'): + pass + for _ in tree.run_kernel(args=kernel_args, build_dir=build_dir, + filter_glob='suite.test2'): + pass + self.assertEqual(kernel_args, ['mem=1G'], + 'run_kernel() should not modify caller args') + self.assertIn('kunit.filter_glob=suite.test1', start_calls[0]) + self.assertIn('kunit.filter_glob=suite.test2', start_calls[1]) + def test_build_reconfig_no_config(self): with tempfile.TemporaryDirectory('') as build_dir: with open(kunit_kernel.get_kunitconfig_path(build_dir), 'w') as f:
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 229/311 accel/amdxdna: Fix NULL pointer dereference of mgmt_chann
In reply to Sasha LevinFrom: Lizhi Hou <[email protected]>
[ Upstream commit 6270ee26e1edd862ea17e3eba148ca8fb2c99dc9 ]
mgmt_chann may be set to NULL if the firmware returns an unexpected error in aie2_send_mgmt_msg_wait(). This can later lead to a NULL pointer dereference in aie2_hw_stop().
Fix this by introducing a dedicated helper to destroy mgmt_chann and by adding proper NULL checks before accessing it.
--- drivers/accel/amdxdna/aie2_message.c | 21 ++++++++++++++++----- drivers/accel/amdxdna/aie2_pci.c | 7 ++----- drivers/accel/amdxdna/aie2_pci.h | 1 + 3 files changed, 19 insertions(+), 10 deletions(-)
drivers/accel/amdxdna/aie2_message.c +16 -5
@@ -40,11 +40,8 @@ static int aie2_send_mgmt_msg_wait(struct amdxdna_dev_hdl *ndev,return -ENODEV ; ret =xdna_send_msg_wait (xdna, ndev->mgmt_chann, msg); -if (ret == -ETIME ) { -xdna_mailbox_stop_channel (ndev->mgmt_chann); -xdna_mailbox_destroy_channel (ndev->mgmt_chann); - ndev->mgmt_chann =NULL ; - } +if (ret == -ETIME ) +aie2_destroy_mgmt_chann (ndev);if (!ret && *hdl->status !=AIE2_STATUS_SUCCESS ) { XDNA_ERR(xdna, "command opcode 0x%x failed, status 0x%x",
@@ -871,6 +868,20 @@ void aie2_msg_init(struct amdxdna_dev_hdl *ndev) ndev->exec_msg_ops = &legacy_exec_message_ops; } +void aie2_destroy_mgmt_chann (struct amdxdna_dev_hdl *ndev) +{ +struct amdxdna_dev *xdna = ndev->xdna; + +drm_WARN_ON (&xdna->ddev, !mutex_is_locked (&xdna->dev_lock)); + +if (!ndev->mgmt_chann) +return ; + +xdna_mailbox_stop_channel (ndev->mgmt_chann); +xdna_mailbox_destroy_channel (ndev->mgmt_chann); + ndev->mgmt_chann =NULL ; +} +static inline struct amdxdna_gem_obj *aie2_cmdlist_get_cmd_buf (struct amdxdna_sched_job *job) {
drivers/accel/amdxdna/aie2_pci.c +2 -5
@@ -343,9 +343,7 @@ static void aie2_hw_stop(struct amdxdna_dev *xdna)aie2_runtime_cfg (ndev,AIE2_RT_CFG_CLK_GATING ,NULL );aie2_mgmt_fw_fini (ndev); -xdna_mailbox_stop_channel (ndev->mgmt_chann); -xdna_mailbox_destroy_channel (ndev->mgmt_chann); - ndev->mgmt_chann =NULL ; +aie2_destroy_mgmt_chann (ndev);drmm_kfree (&xdna->ddev, ndev->mbox); ndev->mbox =NULL ;aie2_psp_stop (ndev->psp_hdl);
@@ -454,8 +452,7 @@ static int aie2_hw_start(struct amdxdna_dev *xdna)return 0 ; destroy_mgmt_chann: -xdna_mailbox_stop_channel (ndev->mgmt_chann); -xdna_mailbox_destroy_channel (ndev->mgmt_chann); +aie2_destroy_mgmt_chann (ndev); stop_psp:aie2_psp_stop (ndev->psp_hdl); fini_smu:
drivers/accel/amdxdna/aie2_pci.h +1 -0
@@ -302,6 +302,7 @@ int aie2_get_array_async_error(struct amdxdna_dev_hdl *ndev,/* aie2_message.c */ void aie2_msg_init (struct amdxdna_dev_hdl *ndev); +void aie2_destroy_mgmt_chann (struct amdxdna_dev_hdl *ndev);int aie2_suspend_fw (struct amdxdna_dev_hdl *ndev);int aie2_resume_fw (struct amdxdna_dev_hdl *ndev);int aie2_set_runtime_cfg (struct amdxdna_dev_hdl *ndev,u32 type,u64 value);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 230/311 drm/amd/display: Use mpc.preblend flag to indicate 3D LUT
In reply to Sasha LevinFrom: Alex Hung <[email protected]>
[ Upstream commit c28b3ec3ca034fd1abc832fef46ce36eb13f8fad ]
[WHAT]
New ASIC's 3D LUT is indicated by mpc.preblend.
--- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c | 6 ++++-- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_colorop.c | 3 ++- 2 files changed, 6 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c +4 -2
@@ -1706,6 +1706,7 @@ __set_dm_plane_colorop_3dlut(struct drm_plane_state *plane_state,struct dc_transfer_func *tf = &dc_plane_state->in_shaper_func;struct drm_atomic_state *state = plane_state->state;const struct amdgpu_device *adev =drm_to_adev (colorop->dev); +bool has_3dlut = adev->dm.dc->caps.color.dpp.hw_3d_lut || adev->dm.dc->caps.color.mpc.preblend;const struct drm_device *dev = colorop->dev;const struct drm_color_lut32 *lut3d;uint32_t lut3d_size;
@@ -1722,7 +1723,7 @@ __set_dm_plane_colorop_3dlut(struct drm_plane_state *plane_state, }if (colorop_state && !colorop_state->bypass && colorop->type ==DRM_COLOROP_3D_LUT ) { -if (!adev->dm.dc->caps.color.dpp.hw_3d_lut) { +if (!has_3dlut) {drm_dbg (dev,"3D LUT is not supported by hardware\n" );return -EINVAL ; }
@@ -1875,6 +1876,7 @@ amdgpu_dm_plane_set_colorop_properties(struct drm_plane_state *plane_state,struct drm_colorop *colorop = plane_state->color_pipeline;struct drm_device *dev = plane_state->plane->dev;struct amdgpu_device *adev =drm_to_adev (dev); +bool has_3dlut = adev->dm.dc->caps.color.dpp.hw_3d_lut || adev->dm.dc->caps.color.mpc.preblend;int ret;/* 1D Curve - DEGAM TF */
@@ -1907,7 +1909,7 @@ amdgpu_dm_plane_set_colorop_properties(struct drm_plane_state *plane_state,if (ret)return ret; -if (adev->dm.dc->caps.color.dpp.hw_3d_lut) { +if (has_3dlut) {/* 1D Curve & LUT - SHAPER TF & LUT */ colorop = colorop->next;if (!colorop) {
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_colorop.c +2 -1
@@ -60,6 +60,7 @@ int amdgpu_dm_initialize_default_pipeline(struct drm_plane *plane, struct drm_prstruct drm_colorop *ops[MAX_COLOR_PIPELINE_OPS ];struct drm_device *dev = plane->dev;struct amdgpu_device *adev =drm_to_adev (dev); +bool has_3dlut = adev->dm.dc->caps.color.dpp.hw_3d_lut || adev->dm.dc->caps.color.mpc.preblend;int ret;int i =0 ;
@@ -112,7 +113,7 @@ int amdgpu_dm_initialize_default_pipeline(struct drm_plane *plane, struct drm_pr i++; -if (adev->dm.dc->caps.color.dpp.hw_3d_lut) { +if (has_3dlut) {/* 1D curve - SHAPER TF */ ops[i] =kzalloc (sizeof (*ops[0 ]),GFP_KERNEL );if (!ops[i]) {
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 231/311 drm/amd/display: Enable DEGAMMA and reject COLOR_PIPELINE+DEGAMMA_LUT
In reply to Sasha LevinFrom: Alex Hung <[email protected]>
[ Upstream commit a4fa2355e0add57253468ef13bd08f11285f3b6e ]
[WHAT]
Create DEGAMMA properties even if color pipeline is enabled, and enforce the mutual exclusion in atomic check by rejecting any commit that attempts to enable both COLOR_PIPELINE on the plane and DEGAMMA_LUT on the CRTC simultaneously.
--- .../drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c | 16 ++++++++-------- .../drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c | 8 ++++++++ 2 files changed, 16 insertions(+), 8 deletions(-)
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c +8 -8
@@ -765,15 +765,15 @@ int amdgpu_dm_crtc_init(struct amdgpu_display_manager *dm, dm->adev->mode_info.crtcs[crtc_index] = acrtc;/* Don't enable DRM CRTC degamma property for -* 1. Degamma is replaced by color pipeline. -* 2. DCE since it doesn't support programmable degamma anywhere. -* 3. DCN401 since pre-blending degamma LUT doesn't apply to cursor. +* 1. DCE since it doesn't support programmable degamma anywhere. +* 2. DCN401 since pre-blending degamma LUT doesn't apply to cursor. +* Note: DEGAMMA properties are created even if the primary plane has the +* COLOR_PIPELINE property. User space can use either the DEGAMMA properties +* or the COLOR_PIPELINE property. An atomic commit which attempts to enable +* both is rejected. */ -if (plane->color_pipeline_property) - has_degamma = false; -else - has_degamma = dm->adev->dm.dc->caps.color.dpp.dcn_arch && - dm->adev->dm.dc->ctx->dce_version !=DCN_VERSION_4_01 ; + has_degamma = dm->adev->dm.dc->caps.color.dpp.dcn_arch && + dm->adev->dm.dc->ctx->dce_version !=DCN_VERSION_4_01 ;drm_crtc_enable_color_mgmt (&acrtc->base, has_degamma ?MAX_COLOR_LUT_ENTRIES :0 , true,MAX_COLOR_LUT_ENTRIES );
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +8 -0
@@ -1256,6 +1256,14 @@ static int amdgpu_dm_plane_atomic_check(struct drm_plane *plane,if (ret)return ret; +/* Reject commits that attempt to use both COLOR_PIPELINE and CRTC DEGAMMA_LUT */ +if (new_plane_state->color_pipeline && new_crtc_state->degamma_lut) { +drm_dbg_atomic (plane->dev, +"[PLANE:%d:%s] COLOR_PIPELINE and CRTC DEGAMMA_LUT cannot be enabled simultaneously\n" , + plane->base.id, plane->name); +return -EINVAL ; + } + ret =amdgpu_dm_plane_fill_dc_scaling_info (adev, new_plane_state, &scaling_info);if (ret)return ret;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 232/311 net: dsa: realtek: rtl8365mb: fix rtl8365mb_phy_ocp_write return value
In reply to Sasha LevinFrom: Mieczyslaw Nalewaj <[email protected]>
[ Upstream commit 7cbe98f7bef965241a5908d50d557008cf998aee ]
Function rtl8365mb_phy_ocp_write() always returns 0, even when an error occurs during register access. This patch fixes the return value to propagate the actual error code from regmap operations.
--- drivers/net/dsa/realtek/rtl8365mb.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/dsa/realtek/rtl8365mb.c +1 -1
@@ -769,7 +769,7 @@ static int rtl8365mb_phy_ocp_write(struct realtek_priv *priv, int phy, out:rtl83xx_unlock (priv); -return 0 ; +return ret; }static int rtl8365mb_phy_read (struct realtek_priv *priv,int phy,int regnum)
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 233/311 bpf/bonding: reject vlan+srcmac xmit_hash_policy change when XDP is loaded
In reply to Sasha LevinFrom: Jiayuan Chen <[email protected]>
[ Upstream commit 479d589b40b836442bbdadc3fdb37f001bb67f26 ]
bond_option_mode_set() already rejects mode changes that would make a loaded XDP program incompatible via bond_xdp_check(). However, bond_option_xmit_hash_policy_set() has no such guard.
For 802.3ad and balance-xor modes, bond_xdp_check() returns false when xmit_hash_policy is vlan+srcmac, because the 802.1q payload is usually absent due to hardware offload. This means a user can:
1. Attach a native XDP program to a bond in 802.3ad/balance-xor mode with a compatible xmit_hash_policy (e.g. layer2+3).
2. Change xmit_hash_policy to vlan+srcmac while XDP remains loaded.
This leaves bond->xdp_prog set but bond_xdp_check() now returning false for the same device. When the bond is later destroyed, dev_xdp_uninstall() calls bond_xdp_set(dev, NULL, NULL) to remove the program, which hits the bond_xdp_check() guard and returns -EOPNOTSUPP, triggering:
WARN_ON(dev_xdp_install(dev, mode, bpf_op, NULL, 0, NULL))
Fix this by rejecting xmit_hash_policy changes to vlan+srcmac when an XDP program is loaded on a bond in 802.3ad or balance-xor mode.
commit 39a0876d595b ("net, bonding: Disallow vlan+srcmac with XDP") introduced bond_xdp_check() which returns false for 802.3ad/balance-xor modes when xmit_hash_policy is vlan+srcmac. The check was wired into bond_xdp_set() to reject XDP attachment with an incompatible policy, but the symmetric path -- preventing xmit_hash_policy from being changed to an incompatible value after XDP is already loaded -- was left unguarded in bond_option_xmit_hash_policy_set().
Note:
commit 094ee6017ea0 ("bonding: check xdp prog when set bond mode") later added a similar guard to bond_option_mode_set(), but bond_option_xmit_hash_policy_set() remained unprotected.
--- drivers/net/bonding/bond_main.c | 9 +++++++-- drivers/net/bonding/bond_options.c | 2 ++ include/net/bonding.h | 1 + 3 files changed, 10 insertions(+), 2 deletions(-)
drivers/net/bonding/bond_main.c +7 -2
@@ -324,7 +324,7 @@ static bool bond_sk_check(struct bonding *bond) } } -bool bond_xdp_check (struct bonding *bond,int mode) +bool __bond_xdp_check (int mode,int xmit_policy) {switch (mode) {case BOND_MODE_ROUNDROBIN :
@@ -335,7 +335,7 @@ bool bond_xdp_check(struct bonding *bond, int mode)/* vlan+srcmac is not supported with XDP as in most cases the 802.1q * payload is not in the packet due to hardware offload. */ -if (bond->params.xmit_policy !=BOND_XMIT_POLICY_VLAN_SRCMAC ) +if (xmit_policy !=BOND_XMIT_POLICY_VLAN_SRCMAC )return true; fallthrough;default :
@@ -343,6 +343,11 @@ bool bond_xdp_check(struct bonding *bond, int mode) } } +bool bond_xdp_check (struct bonding *bond,int mode) +{ +return __bond_xdp_check (mode, bond->params.xmit_policy); +} +/*---------------------------------- VLAN -----------------------------------*/ /* In the following 2 functions, bond_vlan_rx_add_vid and bond_vlan_rx_kill_vid,
drivers/net/bonding/bond_options.c +2 -0
@@ -1574,6 +1574,8 @@ static int bond_option_fail_over_mac_set(struct bonding *bond,static int bond_option_xmit_hash_policy_set (struct bonding *bond,const struct bond_opt_value *newval) { +if (bond->xdp_prog && !__bond_xdp_check (BOND_MODE (bond), newval->value)) +return -EOPNOTSUPP ;netdev_dbg (bond->dev,"Setting xmit hash policy to %s (%llu)\n" , newval->string, newval->value); bond->params.xmit_policy = newval->value;
include/net/bonding.h +1 -0
@@ -698,6 +698,7 @@ void bond_debug_register(struct bonding *bond);void bond_debug_unregister (struct bonding *bond);void bond_debug_reregister (struct bonding *bond);const char *bond_mode_name (int mode); +bool __bond_xdp_check (int mode,int xmit_policy);bool bond_xdp_check (struct bonding *bond,int mode);void bond_setup (struct net_device *bond_dev);unsigned int bond_get_num_tx_queues (void );
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 234/311 octeon_ep: Relocate counter updates before NAPI
In reply to Sasha LevinFrom: Vimlesh Kumar <[email protected]>
[ Upstream commit 18c04a808c436d629d5812ce883e3822a5f5a47f ]
Relocate IQ/OQ IN/OUT_CNTS updates to occur before NAPI completion, and replace napi_complete with napi_complete_done.
Moving the IQ/OQ counter updates before napi_complete_done ensures
1. Counter registers are updated before re-enabling interrupts.
2. Prevents a race where new packets arrive but counters aren't properly synchronized.napi_complete_done (vs napi_complete) allows for better interrupt coalescing.
--- .../ethernet/marvell/octeon_ep/octep_main.c | 19 +++++++++++++++---- 1 file changed, 15 insertions(+), 4 deletions(-)
drivers/net/ethernet/marvell/octeon_ep/octep_main.c +15 -4
@@ -555,12 +555,12 @@ static void octep_clean_irqs(struct octep_device *oct) }/** -* octep_enable_ioq_irq() - Enable MSI-x interrupt of a Tx/Rx queue. +* octep_update_pkt() - Update IQ/OQ IN/OUT_CNT registers. * * @iq: Octeon Tx queue data structure. * @oq: Octeon Rx queue data structure. */ -static void octep_enable_ioq_irq (struct octep_iq *iq,struct octep_oq *oq) +static void octep_update_pkt (struct octep_iq *iq,struct octep_oq *oq) {u32 pkts_pend = oq->pkts_pending;
@@ -576,7 +576,17 @@ static void octep_enable_ioq_irq(struct octep_iq *iq, struct octep_oq *oq) }/* Flush the previous wrties before writing to RESEND bit */ -wmb (); +smp_wmb (); +} + +/** +* octep_enable_ioq_irq() - Enable MSI-x interrupt of a Tx/Rx queue. +* +* @iq: Octeon Tx queue data structure. +* @oq: Octeon Rx queue data structure. +*/ +static void octep_enable_ioq_irq (struct octep_iq *iq,struct octep_oq *oq) +{writeq (1UL <<OCTEP_OQ_INTR_RESEND_BIT , oq->pkts_sent_reg);writeq (1UL <<OCTEP_IQ_INTR_RESEND_BIT , iq->inst_cnt_reg); }
@@ -602,7 +612,8 @@ static int octep_napi_poll(struct napi_struct *napi, int budget)if (tx_pending || rx_done >= budget)return budget; -napi_complete (napi); +octep_update_pkt (ioq_vector->iq, ioq_vector->oq); +napi_complete_done (napi, rx_done);octep_enable_ioq_irq (ioq_vector->iq, ioq_vector->oq);return rx_done; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 235/311 octeon_ep: avoid compiler and IQ/OQ reordering
In reply to Sasha LevinFrom: Vimlesh Kumar <[email protected]>
[ Upstream commit 43b3160cb639079a15daeb5f080120afbfbfc918 ]
Utilize READ_ONCE and WRITE_ONCE APIs for IO queue Tx/Rx variable access to prevent compiler optimization and reordering. Additionally, ensure IO queue OUT/IN_CNT registers are flushed by performing a read-back after writing.
The compiler could reorder reads/writes to pkts_pending, last_pkt_count, etc., causing stale values to be used when calculating packets to process or register updates to send to hardware. The Octeon hardware requires a read-back after writing to OUT_CNT/IN_CNT registers to ensure the write has been flushed through any posted write buffers before the interrupt resend bit is set. Without this, we have observed cases where the hardware didn't properly update its internal state.
wmb/rmb only provides ordering guarantees but doesn't prevent the compiler from performing optimizations like caching in registers, load tearing etc.
--- .../ethernet/marvell/octeon_ep/octep_main.c | 21 +++++++++------ .../net/ethernet/marvell/octeon_ep/octep_rx.c | 27 +++++++++++++------ 2 files changed, 32 insertions(+), 16 deletions(-)
drivers/net/ethernet/marvell/octeon_ep/octep_main.c +13 -8
@@ -562,17 +562,22 @@ static void octep_clean_irqs(struct octep_device *oct)*/ static void octep_update_pkt (struct octep_iq *iq,struct octep_oq *oq) { -u32 pkts_pend = oq->pkts_pending; +u32 pkts_pend =READ_ONCE (oq->pkts_pending); +u32 last_pkt_count =READ_ONCE (oq->last_pkt_count); +u32 pkts_processed =READ_ONCE (iq->pkts_processed); +u32 pkt_in_done =READ_ONCE (iq->pkt_in_done);netdev_dbg (iq->netdev,"enabling intr for Q-%u\n" , iq->q_no); -if (iq->pkts_processed) { -writel (iq->pkts_processed, iq->inst_cnt_reg); - iq->pkt_in_done -= iq->pkts_processed; - iq->pkts_processed =0 ; +if (pkts_processed) { +writel (pkts_processed, iq->inst_cnt_reg); +readl (iq->inst_cnt_reg); +WRITE_ONCE (iq->pkt_in_done, (pkt_in_done - pkts_processed)); +WRITE_ONCE (iq->pkts_processed,0 ); } -if (oq->last_pkt_count - pkts_pend) { -writel (oq->last_pkt_count - pkts_pend, oq->pkts_sent_reg); - oq->last_pkt_count = pkts_pend; +if (last_pkt_count - pkts_pend) { +writel (last_pkt_count - pkts_pend, oq->pkts_sent_reg); +readl (oq->pkts_sent_reg); +WRITE_ONCE (oq->last_pkt_count, pkts_pend); }/* Flush the previous wrties before writing to RESEND bit */
drivers/net/ethernet/marvell/octeon_ep/octep_rx.c +19 -8
@@ -324,10 +324,16 @@ static int octep_oq_check_hw_for_pkts(struct octep_device *oct, struct octep_oq *oq) {u32 pkt_count, new_pkts; +u32 last_pkt_count, pkts_pending; pkt_count =readl (oq->pkts_sent_reg); - new_pkts = pkt_count - oq->last_pkt_count; + last_pkt_count =READ_ONCE (oq->last_pkt_count); + new_pkts = pkt_count - last_pkt_count; +if (pkt_count < last_pkt_count) { +dev_err (oq->dev,"OQ-%u pkt_count(%u) < oq->last_pkt_count(%u)\n" , + oq->q_no, pkt_count, last_pkt_count); + }/* Clear the hardware packets counter register if the rx queue is * being processed continuously with-in a single interrupt and * reached half its max value.
@@ -338,8 +344,9 @@ static int octep_oq_check_hw_for_pkts(struct octep_device *oct, pkt_count =readl (oq->pkts_sent_reg); new_pkts += pkt_count; } - oq->last_pkt_count = pkt_count; - oq->pkts_pending += new_pkts; +WRITE_ONCE (oq->last_pkt_count, pkt_count); + pkts_pending =READ_ONCE (oq->pkts_pending); +WRITE_ONCE (oq->pkts_pending, (pkts_pending + new_pkts));return new_pkts; }
@@ -414,7 +421,7 @@ static int __octep_oq_process_rx(struct octep_device *oct,u16 rx_ol_flags;u32 read_idx; - read_idx = oq->host_read_idx; + read_idx =READ_ONCE (oq->host_read_idx); rx_bytes =0 ; desc_used =0 ;for (pkt =0 ; pkt < pkts_to_process; pkt++) {
@@ -499,7 +506,7 @@ static int __octep_oq_process_rx(struct octep_device *oct,napi_gro_receive (oq->napi, skb); } - oq->host_read_idx = read_idx; +WRITE_ONCE (oq->host_read_idx, read_idx); oq->refill_count += desc_used; oq->stats->packets += pkt; oq->stats->bytes += rx_bytes;
@@ -522,22 +529,26 @@ int octep_oq_process_rx(struct octep_oq *oq, int budget) {u32 pkts_available, pkts_processed, total_pkts_processed;struct octep_device *oct = oq->octep_dev; +u32 pkts_pending; pkts_available =0 ; pkts_processed =0 ; total_pkts_processed =0 ;while (total_pkts_processed < budget) {/* update pending count only when current one exhausted */ -if (oq->pkts_pending ==0 ) + pkts_pending =READ_ONCE (oq->pkts_pending); +if (pkts_pending ==0 )octep_oq_check_hw_for_pkts (oct, oq); + pkts_pending =READ_ONCE (oq->pkts_pending); pkts_available =min (budget - total_pkts_processed, - oq->pkts_pending); + pkts_pending);if (!pkts_available)break ; pkts_processed =__octep_oq_process_rx (oct, oq, pkts_available); - oq->pkts_pending -= pkts_processed; + pkts_pending =READ_ONCE (oq->pkts_pending); +WRITE_ONCE (oq->pkts_pending, (pkts_pending - pkts_processed)); total_pkts_processed += pkts_processed; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 236/311 octeon_ep_vf: Relocate counter updates before NAPI
In reply to Sasha LevinFrom: Vimlesh Kumar <[email protected]>
[ Upstream commit 2ae7d20fb24f598f60faa8f6ecc856dac782261a ]
Relocate IQ/OQ IN/OUT_CNTS updates to occur before NAPI completion. Moving the IQ/OQ counter updates before napi_complete_done ensures
1. Counter registers are updated before re-enabling interrupts.
2. Prevents a race where new packets arrive but counters aren't properly synchronized.
--- .../marvell/octeon_ep_vf/octep_vf_main.c | 17 +++++++++++++++-- 1 file changed, 15 insertions(+), 2 deletions(-)
drivers/net/ethernet/marvell/octeon_ep_vf/octep_vf_main.c +15 -2
@@ -286,12 +286,13 @@ static void octep_vf_clean_irqs(struct octep_vf_device *oct) }/** -* octep_vf_enable_ioq_irq() - Enable MSI-x interrupt of a Tx/Rx queue. +* octep_vf_update_pkt() - Update IQ/OQ IN/OUT_CNT registers. * * @iq: Octeon Tx queue data structure. * @oq: Octeon Rx queue data structure. */ -static void octep_vf_enable_ioq_irq (struct octep_vf_iq *iq,struct octep_vf_oq *oq) + +static void octep_vf_update_pkt (struct octep_vf_iq *iq,struct octep_vf_oq *oq) {u32 pkts_pend = oq->pkts_pending;
@@ -308,6 +309,17 @@ static void octep_vf_enable_ioq_irq(struct octep_vf_iq *iq, struct octep_vf_oq */* Flush the previous wrties before writing to RESEND bit */ smp_wmb (); +} + +/** +* octep_vf_enable_ioq_irq() - Enable MSI-x interrupt of a Tx/Rx queue. +* +* @iq: Octeon Tx queue data structure. +* @oq: Octeon Rx queue data structure. +*/ +static void octep_vf_enable_ioq_irq (struct octep_vf_iq *iq, +struct octep_vf_oq *oq) +{writeq (1UL <<OCTEP_VF_OQ_INTR_RESEND_BIT , oq->pkts_sent_reg);writeq (1UL <<OCTEP_VF_IQ_INTR_RESEND_BIT , iq->inst_cnt_reg); }
@@ -333,6 +345,7 @@ static int octep_vf_napi_poll(struct napi_struct *napi, int budget)if (tx_pending || rx_done >= budget)return budget; +octep_vf_update_pkt (ioq_vector->iq, ioq_vector->oq);if (likely (napi_complete_done (napi, rx_done)))octep_vf_enable_ioq_irq (ioq_vector->iq, ioq_vector->oq);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 237/311 octeon_ep_vf: avoid compiler and IQ/OQ reordering
In reply to Sasha LevinFrom: Vimlesh Kumar <[email protected]>
[ Upstream commit 6c73126ecd1080351b468fe43353b2f705487f44 ]
Utilize READ_ONCE and WRITE_ONCE APIs for IO queue Tx/Rx variable access to prevent compiler optimization and reordering. Additionally, ensure IO queue OUT/IN_CNT registers are flushed by performing a read-back after writing.
The compiler could reorder reads/writes to pkts_pending, last_pkt_count, etc., causing stale values to be used when calculating packets to process or register updates to send to hardware. The Octeon hardware requires a read-back after writing to OUT_CNT/IN_CNT registers to ensure the write has been flushed through any posted write buffers before the interrupt resend bit is set. Without this, we have observed cases where the hardware didn't properly update its internal state.
wmb/rmb only provides ordering guarantees but doesn't prevent the compiler from performing optimizations like caching in registers, load tearing etc.
--- .../marvell/octeon_ep_vf/octep_vf_main.c | 21 ++++++++------ .../marvell/octeon_ep_vf/octep_vf_rx.c | 28 +++++++++++++------ 2 files changed, 33 insertions(+), 16 deletions(-)
drivers/net/ethernet/marvell/octeon_ep_vf/octep_vf_main.c +13 -8
@@ -294,17 +294,22 @@ static void octep_vf_clean_irqs(struct octep_vf_device *oct)static void octep_vf_update_pkt (struct octep_vf_iq *iq,struct octep_vf_oq *oq) { -u32 pkts_pend = oq->pkts_pending; +u32 pkts_pend =READ_ONCE (oq->pkts_pending); +u32 last_pkt_count =READ_ONCE (oq->last_pkt_count); +u32 pkts_processed =READ_ONCE (iq->pkts_processed); +u32 pkt_in_done =READ_ONCE (iq->pkt_in_done);netdev_dbg (iq->netdev,"enabling intr for Q-%u\n" , iq->q_no); -if (iq->pkts_processed) { -writel (iq->pkts_processed, iq->inst_cnt_reg); - iq->pkt_in_done -= iq->pkts_processed; - iq->pkts_processed =0 ; +if (pkts_processed) { +writel (pkts_processed, iq->inst_cnt_reg); +readl (iq->inst_cnt_reg); +WRITE_ONCE (iq->pkt_in_done, (pkt_in_done - pkts_processed)); +WRITE_ONCE (iq->pkts_processed,0 ); } -if (oq->last_pkt_count - pkts_pend) { -writel (oq->last_pkt_count - pkts_pend, oq->pkts_sent_reg); - oq->last_pkt_count = pkts_pend; +if (last_pkt_count - pkts_pend) { +writel (last_pkt_count - pkts_pend, oq->pkts_sent_reg); +readl (oq->pkts_sent_reg); +WRITE_ONCE (oq->last_pkt_count, pkts_pend); }/* Flush the previous wrties before writing to RESEND bit */
drivers/net/ethernet/marvell/octeon_ep_vf/octep_vf_rx.c +20 -8
@@ -325,9 +325,16 @@ static int octep_vf_oq_check_hw_for_pkts(struct octep_vf_device *oct, struct octep_vf_oq *oq) {u32 pkt_count, new_pkts; +u32 last_pkt_count, pkts_pending; pkt_count =readl (oq->pkts_sent_reg); - new_pkts = pkt_count - oq->last_pkt_count; + last_pkt_count =READ_ONCE (oq->last_pkt_count); + new_pkts = pkt_count - last_pkt_count; + +if (pkt_count < last_pkt_count) { +dev_err (oq->dev,"OQ-%u pkt_count(%u) < oq->last_pkt_count(%u)\n" , + oq->q_no, pkt_count, last_pkt_count); + }/* Clear the hardware packets counter register if the rx queue is * being processed continuously with-in a single interrupt and
@@ -339,8 +346,9 @@ static int octep_vf_oq_check_hw_for_pkts(struct octep_vf_device *oct, pkt_count =readl (oq->pkts_sent_reg); new_pkts += pkt_count; } - oq->last_pkt_count = pkt_count; - oq->pkts_pending += new_pkts; +WRITE_ONCE (oq->last_pkt_count, pkt_count); + pkts_pending =READ_ONCE (oq->pkts_pending); +WRITE_ONCE (oq->pkts_pending, (pkts_pending + new_pkts));return new_pkts; }
@@ -369,7 +377,7 @@ static int __octep_vf_oq_process_rx(struct octep_vf_device *oct,struct sk_buff *skb;u32 read_idx; - read_idx = oq->host_read_idx; + read_idx =READ_ONCE (oq->host_read_idx); rx_bytes =0 ; desc_used =0 ;for (pkt =0 ; pkt < pkts_to_process; pkt++) {
@@ -463,7 +471,7 @@ static int __octep_vf_oq_process_rx(struct octep_vf_device *oct,napi_gro_receive (oq->napi, skb); } - oq->host_read_idx = read_idx; +WRITE_ONCE (oq->host_read_idx, read_idx); oq->refill_count += desc_used; oq->stats->packets += pkt; oq->stats->bytes += rx_bytes;
@@ -486,22 +494,26 @@ int octep_vf_oq_process_rx(struct octep_vf_oq *oq, int budget) {u32 pkts_available, pkts_processed, total_pkts_processed;struct octep_vf_device *oct = oq->octep_vf_dev; +u32 pkts_pending; pkts_available =0 ; pkts_processed =0 ; total_pkts_processed =0 ;while (total_pkts_processed < budget) {/* update pending count only when current one exhausted */ -if (oq->pkts_pending ==0 ) + pkts_pending =READ_ONCE (oq->pkts_pending); +if (pkts_pending ==0 )octep_vf_oq_check_hw_for_pkts (oct, oq); + pkts_pending =READ_ONCE (oq->pkts_pending); pkts_available =min (budget - total_pkts_processed, - oq->pkts_pending); + pkts_pending);if (!pkts_available)break ; pkts_processed =__octep_vf_oq_process_rx (oct, oq, pkts_available); - oq->pkts_pending -= pkts_processed; + pkts_pending =READ_ONCE (oq->pkts_pending); +WRITE_ONCE (oq->pkts_pending, (pkts_pending - pkts_processed)); total_pkts_processed += pkts_processed; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 238/311 wifi: cw1200: Fix locking in error paths
In reply to Sasha LevinFrom: Bart Van Assche <[email protected]>
[ Upstream commit d98c24617a831e92e7224a07dcaed2dd0b02af96 ]
cw1200_wow_suspend() must only return with priv->conf_mutex locked if it returns zero. This mutex must be unlocked if an error is returned. Add mutex_unlock() calls to the error paths from which that call is missing. This has been detected by the Clang thread-safety analyzer.
--- drivers/net/wireless/st/cw1200/pm.c | 2 ++ 1 file changed, 2 insertions(+)
drivers/net/wireless/st/cw1200/pm.c +2 -0
@@ -264,12 +264,14 @@ int cw1200_wow_suspend(struct ieee80211_hw *hw, struct cfg80211_wowlan *wowlan)wiphy_err (priv->hw->wiphy,"PM request failed: %d. WoW is disabled.\n" , ret);cw1200_wow_resume (hw); +mutex_unlock (&priv->conf_mutex);return -EBUSY ; }/* Force resume if event is coming from the device. */ if (atomic_read (&priv->bh_rx)) {cw1200_wow_resume (hw); +mutex_unlock (&priv->conf_mutex);return -EAGAIN ; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 239/311 wifi: wlcore: Fix a locking bug
In reply to Sasha LevinFrom: Bart Van Assche <[email protected]>
[ Upstream commit 72c6df8f284b3a49812ce2ac136727ace70acc7c ]
Make sure that wl->mutex is locked before it is unlocked. This has been detected by the Clang thread-safety analyzer.
--- drivers/net/wireless/ti/wlcore/main.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/net/wireless/ti/wlcore/main.c +2 -2
@@ -1875,6 +1875,8 @@ static int __maybe_unused wl1271_op_resume(struct ieee80211_hw *hw) wl->wow_enabled);WARN_ON (!wl->wow_enabled); +mutex_lock (&wl->mutex); + ret =pm_runtime_force_resume (wl->dev);if (ret <0 ) {wl1271_error ("ELP wakeup failure!" );
@@ -1891,8 +1893,6 @@ static int __maybe_unused wl1271_op_resume(struct ieee80211_hw *hw) run_irq_work = true;spin_unlock_irqrestore (&wl->wl_lock, flags); -mutex_lock (&wl->mutex); -/* test the recovery flag before calling any SDIO functions */ pending_recovery =test_bit (WL1271_FLAG_RECOVERY_IN_PROGRESS , &wl->flags);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 240/311 wifi: mt76: mt7996: Fix possible oob access in mt7996_mac_write_txwi_80211()
In reply to Sasha LevinFrom: Lorenzo Bianconi <[email protected]>
[ Upstream commit 60862846308627e9e15546bb647a00de44deb27b ]
Check frame length before accessing the mgmt fields in mt7996_mac_write_txwi_80211 in order to avoid a possible oob access.
--- drivers/net/wireless/mediatek/mt76/mt7996/mac.c | 1 + 1 file changed, 1 insertion(+)
drivers/net/wireless/mediatek/mt76/mt7996/mac.c +1 -0
@@ -800,6 +800,7 @@ mt7996_mac_write_txwi_80211(struct mt7996_dev *dev, __le32 *txwi,u32 val;if (ieee80211_is_action (fc) && + skb->len >=IEEE80211_MIN_ACTION_SIZE +1 && mgmt->u.action.category ==WLAN_CATEGORY_BACK && mgmt->u.action.u.addba_req.action_code ==WLAN_ACTION_ADDBA_REQ ) {if (is_mt7990 (&dev->mt76))
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 241/311 wifi: mt76: mt7925: Fix possible oob access in mt7925_mac_write_txwi_80211()
In reply to Sasha LevinFrom: Lorenzo Bianconi <[email protected]>
[ Upstream commit c41a9abd6ae31d130e8f332e7c8800c4c866234b ]
Check frame length before accessing the mgmt fields in mt7925_mac_write_txwi_80211 in order to avoid a possible oob access.
--- drivers/net/wireless/mediatek/mt76/mt7925/mac.c | 1 + 1 file changed, 1 insertion(+)
drivers/net/wireless/mediatek/mt76/mt7925/mac.c +1 -0
@@ -668,6 +668,7 @@ mt7925_mac_write_txwi_80211(struct mt76_dev *dev, __le32 *txwi,u32 val;if (ieee80211_is_action (fc) && + skb->len >=IEEE80211_MIN_ACTION_SIZE +1 && mgmt->u.action.category ==WLAN_CATEGORY_BACK && mgmt->u.action.u.addba_req.action_code ==WLAN_ACTION_ADDBA_REQ ) tid =MT_TX_ADDBA ;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 242/311 wifi: mt76: Fix possible oob access in mt76_connac2_mac_write_txwi_80211()
In reply to Sasha LevinFrom: Lorenzo Bianconi <[email protected]>
[ Upstream commit 4e10a730d1b511ff49723371ed6d694dd1b2c785 ]
Check frame length before accessing the mgmt fields in mt76_connac2_mac_write_txwi_80211 in order to avoid a possible oob access.
[fix check to also cover mgmt->u.action.u.addba_req.capab, correct Fixes tag]
--- drivers/net/wireless/mediatek/mt76/mt76_connac_mac.c | 1 + 1 file changed, 1 insertion(+)
drivers/net/wireless/mediatek/mt76/mt76_connac_mac.c +1 -0
@@ -413,6 +413,7 @@ mt76_connac2_mac_write_txwi_80211(struct mt76_dev *dev, __le32 *txwi,u32 val;if (ieee80211_is_action (fc) && + skb->len >=IEEE80211_MIN_ACTION_SIZE +1 +1 +2 && mgmt->u.action.category ==WLAN_CATEGORY_BACK && mgmt->u.action.u.addba_req.action_code ==WLAN_ACTION_ADDBA_REQ ) {u16 capab =le16_to_cpu (mgmt->u.action.u.addba_req.capab);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 243/311 indirect_call_wrapper: do not reevaluate function pointer
In reply to Sasha LevinFrom: Eric Dumazet <[email protected]>
[ Upstream commit 710f5c76580306cdb9ec51fac8fcf6a8faff7821 ]
We have an increasing number of READ_ONCE(xxx->function) combined with INDIRECT_CALL_[1234]() helpers.
Unfortunately this forces INDIRECT_CALL_[1234]() to read xxx->function many times, which is not what we wanted.
Fix these macros so that xxx->function value is not reloaded.
$ scripts/bloat-o-meter -t vmlinux.0 vmlinux add/remove: 0/0 grow/shrink: 1/65 up/down: 122/-1084 (-962) Function old new delta ip_push_pending_frames 59 181 +122 ip6_finish_output 687 681 -6 __udp_enqueue_schedule_skb 1078 1072 -6 ioam6_output 2319 2312 -7 xfrm4_rcv_encap_finish2 64 56 -8 xfrm4_output 297 289 -8 vrf_ip_local_out 278 270 -8 vrf_ip6_local_out 278 270 -8 seg6_input_finish 64 56 -8 rpl_output 700 692 -8 ipmr_forward_finish 124 116 -8 ip_forward_finish 143 135 -8 ip6mr_forward2_finish 100 92 -8 ip6_forward_finish 73 65 -8 input_action_end_bpf 1091 1083 -8 dst_input 52 44 -8 __xfrm6_output 801 793 -8 __xfrm4_output 83 75 -8 bpf_input 500 491 -9 __tcp_check_space 530 521 -9 input_action_end_dt6 291 280 -11 vti6_tnl_xmit 1634 1622 -12 bpf_xmit 1203 1191 -12 rpl_input 497 483 -14 rawv6_send_hdrinc 1355 1341 -14 ndisc_send_skb 1030 1016 -14 ipv6_srh_rcv 1377 1363 -14 ip_send_unicast_reply 1253 1239 -14 ip_rcv_finish 226 212 -14 ip6_rcv_finish 300 286 -14 input_action_end_x_core 205 191 -14 input_action_end_x 355 341 -14 input_action_end_t 205 191 -14 input_action_end_dx6_finish 127 113 -14 input_action_end_dx4_finish 373 359 -14 input_action_end_dt4 426 412 -14 input_action_end_core 186 172 -14 input_action_end_b6_encap 292 278 -14 input_action_end_b6 198 184 -14 igmp6_send 1332 1318 -14 ip_sublist_rcv 864 848 -16 ip6_sublist_rcv 1091 1075 -16 ipv6_rpl_srh_rcv 1937 1920 -17 xfrm_policy_queue_process 1246 1228 -18 seg6_output_core 903 885 -18 mld_sendpack 856 836 -20 NF_HOOK 756 736 -20 vti_tunnel_xmit 1447 1426 -21 input_action_end_dx6 664 642 -22 input_action_end 1502 1480 -22 sock_sendmsg_nosec 134 111 -23 ip6mr_forward2 388 364 -24 sock_recvmsg_nosec 134 109 -25 seg6_input_core 836 810 -26 ip_send_skb 172 146 -26 ip_local_out 140 114 -26 ip6_local_out 140 114 -26 __sock_sendmsg 162 136 -26 __ip_queue_xmit 1196 1170 -26 __ip_finish_output 405 379 -26 ipmr_queue_fwd_xmit 373 346 -27 sock_recvmsg 173 145 -28 ip6_xmit 1635 1607 -28 xfrm_output_resume 1418 1389 -29 ip_build_and_send_pkt 625 591 -34 dst_output 504 432 -72 Total: Before=25217686, After=25216724, chg -0.00%
--- include/linux/indirect_call_wrapper.h | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-)
include/linux/indirect_call_wrapper.h +11 -7
@@ -16,22 +16,26 @@*/ #define INDIRECT_CALL_1 (f, f1, ...) \ ({ \ - likely(f == f1) ? f1(__VA_ARGS__) : f(__VA_ARGS__); \ + typeof(f) __f1 = (f); \ + likely(__f1 == f1) ? f1(__VA_ARGS__) : __f1(__VA_ARGS__); \ })#define INDIRECT_CALL_2 (f, f2, f1, ...) \ ({ \ - likely(f == f2) ? f2(__VA_ARGS__) : \ - INDIRECT_CALL_1(f, f1, __VA_ARGS__); \ + typeof(f) __f2 = (f); \ + likely(__f2 == f2) ? f2(__VA_ARGS__) : \ + INDIRECT_CALL_1(__f2, f1, __VA_ARGS__); \ })#define INDIRECT_CALL_3 (f, f3, f2, f1, ...) \ ({ \ - likely(f == f3) ? f3(__VA_ARGS__) : \ - INDIRECT_CALL_2(f, f2, f1, __VA_ARGS__); \ + typeof(f) __f3 = (f); \ + likely(__f3 == f3) ? f3(__VA_ARGS__) : \ + INDIRECT_CALL_2(__f3, f2, f1, __VA_ARGS__); \ })#define INDIRECT_CALL_4 (f, f4, f3, f2, f1, ...) \ ({ \ - likely(f == f4) ? f4(__VA_ARGS__) : \ - INDIRECT_CALL_3(f, f3, f2, f1, __VA_ARGS__); \ + typeof(f) __f4 = (f); \ + likely(__f4 == f4) ? f4(__VA_ARGS__) : \ + INDIRECT_CALL_3(__f4, f3, f2, f1, __VA_ARGS__); \ })#define INDIRECT_CALLABLE_DECLARE (f) f
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 244/311 net/rds: Fix circular locking dependency in rds_tcp_tune
In reply to Sasha LevinFrom: Allison Henderson <[email protected]>
[ Upstream commit 6a877ececd6daa002a9a0002cd0fbca6592a9244 ]
syzbot reported a circular locking dependency in rds_tcp_tune() where sk_net_refcnt_upgrade() is called while holding the socket lock:
======================================================
WARNING: possible circular locking dependency detected
======================================================
kworker/u10:8/15040 is trying to acquire lock:ffffffff8e9aaf80 (fs_reclaim){+.+.}-{0:0},
at: __kmalloc_cache_noprof+0x4b/0x6f0
but task is already holding lock:ffff88805a3c1ce0 (k-sk_lock-AF_INET6){+.+.}-{0:0},
at: rds_tcp_tune+0xd7/0x930
The issue occurs because sk_net_refcnt_upgrade() performs memory allocation (via get_net_track() -> ref_tracker_alloc()) while the socket lock is held, creating a circular dependency with fs_reclaim.
Fix this by moving sk_net_refcnt_upgrade() outside the socket lock critical section. This is safe because the fields modified by the sk_net_refcnt_upgrade() call (sk_net_refcnt, ns_tracker) are not accessed by any concurrent code path at this point.
v2:
- Corrected fixes tag
- check patch line wrap nits
- ai commentary nits
--- net/rds/tcp.c | 14 ++++++++++---- 1 file changed, 10 insertions(+), 4 deletions(-)
net/rds/tcp.c +10 -4
@@ -495,18 +495,24 @@ bool rds_tcp_tune(struct socket *sock)struct rds_tcp_net *rtn;tcp_sock_set_nodelay (sock->sk); -lock_sock (sk);/* TCP timer functions might access net namespace even after * a process which created this net namespace terminated. */ if (!sk->sk_net_refcnt) { -if (!maybe_get_net (net)) { -release_sock (sk); +if (!maybe_get_net (net))return false; - } +/* +* sk_net_refcnt_upgrade() must be called before lock_sock() +* because it does a GFP_KERNEL allocation, which can trigger +* fs_reclaim and create a circular lock dependency with the +* socket lock. The fields it modifies (sk_net_refcnt, +* ns_tracker) are not accessed by any concurrent code path +* at this point. +*/ sk_net_refcnt_upgrade (sk);put_net (net); } +lock_sock (sk); rtn =net_generic (net, rds_tcp_netid);if (rtn->sndbuf_size >0 ) { sk->sk_sndbuf = rtn->sndbuf_size;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 245/311 xen/acpi-processor: fix _CST detection using undersized evaluation buffer
In reply to Sasha LevinFrom: David Thomson <[email protected]>
[ Upstream commit 8b57227d59a86fc06d4f09de08f98133680f2cae ]
read_acpi_id() attempts to evaluate _CST using a stack buffer of sizeof(union acpi_object) (48 bytes), but _CST returns a nested Package of sub-Packages (one per C-state, each containing a register descriptor, type, latency, and power) requiring hundreds of bytes. The evaluation always fails with AE_BUFFER_OVERFLOW.
On modern systems using FFH/MWAIT entry (where pblk is zero), this causes the function to return before setting the acpi_id_cst_present bit. In check_acpi_ids(), flags.power is then zero for all Phase 2 CPUs (physical CPUs beyond dom0's vCPU count), so push_cxx_to_hypervisor() is never called for them.
On a system with dom0_max_vcpus=2 and 8 physical CPUs, only PCPUs 0-1 receive C-state data. PCPUs 2-7 are stuck in C0/C1 idle, unable to enter C2/C3. This costs measurable wall power (4W observed on an Intel Core Ultra 7 265K with Xen 4.20).
The function never uses the _CST return value -- it only needs to know whether _CST exists. Replace the broken acpi_evaluate_object() call with acpi_has_method(), which correctly detects _CST presence using acpi_get_handle() without any buffer allocation. This brings C-state detection to parity with the P-state path, which already works correctly for Phase 2 CPUs.
Message-ID: <[email protected]>
--- drivers/xen/xen-acpi-processor.c | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-)
drivers/xen/xen-acpi-processor.c +2 -5
@@ -379,11 +379,8 @@ read_acpi_id(acpi_handle handle, u32 lvl, void *context, void **rv) acpi_psd[acpi_id].domain); } - status =acpi_evaluate_object (handle,"_CST" ,NULL , &buffer); -if (ACPI_FAILURE (status)) { -if (!pblk) -return AE_OK ; - } +if (!pblk && !acpi_has_method (handle,"_CST" )) +return AE_OK ;/* .. and it has a C-state */ __set_bit (acpi_id, acpi_id_cst_present);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 246/311 ASoC: SDCA: Add allocation failure check for Entity name
In reply to Sasha LevinFrom: Charles Keepax <[email protected]>
[ Upstream commit 27990181031fdcdbe0f7c46011f6404e5d116386 ]
Currently find_sdca_entity_iot() can allocate a string for the Entity name but it doesn't check if that allocation succeeded. Add the missing NULL check after the allocation.
--- sound/soc/sdca/sdca_functions.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)
sound/soc/sdca/sdca_functions.c +4 -1
@@ -1120,9 +1120,12 @@ static int find_sdca_entity_iot(struct device *dev,if (!terminal->is_dataport) {const char *type_name =sdca_find_terminal_name (terminal->type); -if (type_name) +if (type_name) { entity->label =devm_kasprintf (dev,GFP_KERNEL ,"%s %s" , entity->label, type_name); +if (!entity->label) +return -ENOMEM ; + } } ret = fwnode_property_read_u32(entity_node,
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 247/311 ice: fix adding AQ LLDP filter for VF
In reply to Sasha LevinFrom: Larysa Zaremba <[email protected]>
[ Upstream commit eef33aa44935d001747ca97703c08dd6f9031162 ]
The referenced commit came from a misunderstanding of the FW LLDP filter AQ (Admin Queue) command due to the error in the internal documentation. Contrary to the assumptions in the original commit, VFs can be added and deleted from this filter without any problems. Introduced dev_info message proved to be useful, so reverting the whole commit does not make sense.
Without this fix, trusted VFs do not receive LLDP traffic, if there is an AQ LLDP filter on PF. When trusted VF attempts to add an LLDP multicast MAC address, the following message can be seen in dmesg on host:
ice 0000:33:00.0: Failed to add Rx LLDP rule on VSI 20 error: -95
Revert checking VSI type when adding LLDP filter through AQ.
--- drivers/net/ethernet/intel/ice/ice_common.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/intel/ice/ice_common.c +1 -1
@@ -6429,7 +6429,7 @@ int ice_lldp_fltr_add_remove(struct ice_hw *hw, struct ice_vsi *vsi, bool add)struct ice_aqc_lldp_filter_ctrl *cmd;struct libie_aq_desc desc; -if (vsi->type !=ICE_VSI_PF || !ice_fw_supports_lldp_fltr_ctrl (hw)) +if (!ice_fw_supports_lldp_fltr_ctrl (hw))return -EOPNOTSUPP ; cmd =libie_aq_raw (&desc);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 248/311 ice: Fix memory leak in ice_set_ringparam()
In reply to Sasha LevinFrom: Zilin Guan <[email protected]>
[ Upstream commit fe868b499d16f55bbeea89992edb98043c9de416 ]
In ice_set_ringparam, tx_rings and xdp_rings are allocated before rx_rings. If the allocation of rx_rings fails, the code jumps to the done label leaking both tx_rings and xdp_rings. Furthermore, if the setup of an individual Rx ring fails during the loop, the code jumps to the free_tx label which releases tx_rings but leaks xdp_rings.
Fix this by introducing a free_xdp label and updating the error paths to ensure both xdp_rings and tx_rings are properly freed if rx_rings allocation or setup fails.
Compile tested only. Issue found using a prototype static analysis tool and code review.
--- drivers/net/ethernet/intel/ice/ice_ethtool.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-)
drivers/net/ethernet/intel/ice/ice_ethtool.c +9 -2
@@ -3322,7 +3322,7 @@ ice_set_ringparam(struct net_device *netdev, struct ethtool_ringparam *ring, rx_rings =kcalloc (vsi->num_rxq,sizeof (*rx_rings),GFP_KERNEL );if (!rx_rings) { err = -ENOMEM ; - goto done; + goto free_xdp; }ice_for_each_rxq (vsi, i) {
@@ -3349,7 +3349,7 @@ ice_set_ringparam(struct net_device *netdev, struct ethtool_ringparam *ring, }kfree (rx_rings); err = -ENOMEM ; - goto free_tx; + goto free_xdp; } }
@@ -3402,6 +3402,13 @@ ice_set_ringparam(struct net_device *netdev, struct ethtool_ringparam *ring, } goto done; +free_xdp: +if (xdp_rings) { +ice_for_each_xdp_txq (vsi, i) +ice_free_tx_ring (&xdp_rings[i]); +kfree (xdp_rings); + } + free_tx:/* error cleanup if the Rx allocations failed after getting Tx */ if (tx_rings) {
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 249/311 libie: don't unroll if fwlog isn't supported
In reply to Sasha LevinFrom: Michal Swiatkowski <[email protected]>
[ Upstream commit 636cc3bd12f499c74eaf5dc9a7d5b832f1bb24ed ]
The libie_fwlog_deinit() function can be called during driver unload even when firmware logging was never properly initialized. This led to call trace:
[ 148.576156] Oops: Oops: 0000 [#1] SMP NOPTI [ 148.576167] CPU: 80 UID: 0 PID: 12843 Comm: rmmod Kdump: loaded Not tainted 6.17.0-rc7next-queue-3oct-01915-g06d79d51cf51 #1 PREEMPT(full) [ 148.576177] Hardware name: HPE ProLiant DL385 Gen10 Plus/ProLiant DL385 Gen10 Plus, BIOS A42 07/18/2020 [ 148.576182] RIP: 0010:__dev_printk+0x16/0x70 [ 148.576196] Code: 1f 44 00 00 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 0f 1f 44 00 00 41 55 41 54 49 89 d4 55 48 89 fd 53 48 85 f6 74 3c <4c> 8b 6e 50 48 89 f3 4d 85 ed 75 03 4c 8b 2e 48 89 df e8 f3 27 98 [ 148.576204] RSP: 0018:ffffd2fd7ea17a48 EFLAGS: 00010202 [ 148.576211] RAX: ffffd2fd7ea17aa0 RBX: ffff8eb288ae2000 RCX: 0000000000000000 [ 148.576217] RDX: ffffd2fd7ea17a70 RSI: 00000000000000c8 RDI: ffffffffb68d3d88 [ 148.576222] RBP: ffffffffb68d3d88 R08: 0000000000000000 R09: 0000000000000000 [ 148.576227] R10: 00000000000000c8 R11: ffff8eb2b1a49400 R12: ffffd2fd7ea17a70 [ 148.576231] R13: ffff8eb3141fb000 R14: ffffffffc1215b48 R15: ffffffffc1215bd8 [ 148.576236] FS: 00007f5666ba6740(0000) GS:ffff8eb2472b9000(0000) knlGS:0000000000000000 [ 148.576242] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 148.576247] CR2: 0000000000000118 CR3: 000000011ad17000 CR4: 0000000000350ef0 [ 148.576252] Call Trace: [ 148.576258] <TASK> [ 148.576269] _dev_warn+0x7c/0x96 [ 148.576290] libie_fwlog_deinit+0x112/0x117 [libie_fwlog] [ 148.576303] ixgbe_remove+0x63/0x290 [ixgbe] [ 148.576342] pci_device_remove+0x42/0xb0 [ 148.576354] device_release_driver_internal+0x19c/0x200 [ 148.576365] driver_detach+0x48/0x90 [ 148.576372] bus_remove_driver+0x6d/0xf0 [ 148.576383] pci_unregister_driver+0x2e/0xb0 [ 148.576393] ixgbe_exit_module+0x1c/0xd50 [ixgbe] [ 148.576430] __do_sys_delete_module.isra.0+0x1bc/0x2e0 [ 148.576446] do_syscall_64+0x7f/0x980
It can be reproduced by trying to unload ixgbe driver in recovery mode.
Fix that by checking if fwlog is supported before doing unroll.
--- drivers/net/ethernet/intel/libie/fwlog.c | 4 ++++ 1 file changed, 4 insertions(+)
drivers/net/ethernet/intel/libie/fwlog.c +4 -0
@@ -1051,6 +1051,10 @@ void libie_fwlog_deinit(struct libie_fwlog *fwlog) {int status; +/* if FW logging isn't supported it means no configuration was done */ +if (!libie_fwlog_supported (fwlog)) +return ; +/* make sure FW logging is disabled to not put the FW in a weird state * for the next driver load */
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 250/311 iavf: fix netdev->max_mtu to respect actual hardware limit
In reply to Sasha LevinFrom: Kohei Enju <[email protected]>
[ Upstream commit b84852170153671bb0fa6737a6e48370addd8e1a ]
iavf sets LIBIE_MAX_MTU as netdev->max_mtu, ignoring vf_res->max_mtu from PF [1]. This allows setting an MTU beyond the actual hardware limit, causing TX queue timeouts [2].
Set correct netdev->max_mtu using vf_res->max_mtu from the PF.
Note that currently PF drivers such as ice/i40e set the frame size in vf_res->max_mtu, not MTU. Convert vf_res->max_mtu to MTU before setting netdev->max_mtu.
[1]
# ip -j -d link show $DEV | jq '.[0].max_mtu' 16356
[2]
iavf 0000:00:05.0 enp0s5: NETDEV WATCHDOG: CPU: 1: transmit queue 0 timed out 5692 ms iavf 0000:00:05.0 enp0s5: NIC Link is Up Speed is 10 Gbps Full Duplex iavf 0000:00:05.0 enp0s5: NETDEV WATCHDOG: CPU: 6: transmit queue 3 timed out 5312 ms iavf 0000:00:05.0 enp0s5: NIC Link is Up Speed is 10 Gbps Full Duplex
...
--- drivers/net/ethernet/intel/iavf/iavf_main.c | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-)
drivers/net/ethernet/intel/iavf/iavf_main.c +16 -1
@@ -2797,7 +2797,22 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter) netdev->watchdog_timeo =5 *HZ ; netdev->min_mtu =ETH_MIN_MTU ; - netdev->max_mtu =LIBIE_MAX_MTU ; + +/* PF/VF API: vf_res->max_mtu is max frame size (not MTU). +* Convert to MTU. +*/ +if (!adapter->vf_res->max_mtu) { + netdev->max_mtu =LIBIE_MAX_MTU ; + }else if (adapter->vf_res->max_mtu <LIBETH_RX_LL_LEN +ETH_MIN_MTU || + adapter->vf_res->max_mtu > +LIBETH_RX_LL_LEN +LIBIE_MAX_MTU ) { +netdev_warn_once (adapter->netdev, +"invalid max frame size %d from PF, using default MTU %d" , + adapter->vf_res->max_mtu,LIBIE_MAX_MTU ); + netdev->max_mtu =LIBIE_MAX_MTU ; + }else { + netdev->max_mtu = adapter->vf_res->max_mtu -LIBETH_RX_LL_LEN ; + }if (!is_valid_ether_addr (adapter->hw.mac.addr)) { dev_info(&pdev->dev, "Invalid MAC address %pM, using random\n",
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 251/311 igb: Fix trigger of incorrect irq in igb_xsk_wakeup
In reply to Sasha LevinFrom: Vivek Behera <[email protected]>
[ Upstream commit d4c13ab36273a8c318ba06799793cc1f5d9c6fa1 ]
The current implementation in the igb_xsk_wakeup expects the Rx and Tx queues to share the same irq. This would lead to triggering of incorrect irq in split irq configuration. This patch addresses this issue which could impact environments with 2 active cpu cores
or when the number of queues is reduced to 2 or less
cat /proc/interrupts | grep eno2
167: 0 0 0 0 IR-PCI-MSIX-0000:08:00.0
0-edge eno2
168: 0 0 0 0 IR-PCI-MSIX-0000:08:00.0
1-edge eno2-rx-0
169: 0 0 0 0 IR-PCI-MSIX-0000:08:00.0
2-edge eno2-rx-1
170: 0 0 0 0 IR-PCI-MSIX-0000:08:00.0
3-edge eno2-tx-0
171: 0 0 0 0 IR-PCI-MSIX-0000:08:00.0
4-edge eno2-tx-1
Furthermore it uses the flags input argument to trigger either rx, tx or both rx and tx irqs as specified in the ndo_xsk_wakeup api documentation
--- drivers/net/ethernet/intel/igb/igb_xsk.c | 38 +++++++++++++++++++----- 1 file changed, 30 insertions(+), 8 deletions(-)
drivers/net/ethernet/intel/igb/igb_xsk.c +30 -8
@@ -524,6 +524,16 @@ bool igb_xmit_zc(struct igb_ring *tx_ring, struct xsk_buff_pool *xsk_pool)return nb_pkts < budget; } +static u32 igb_sw_irq_prep (struct igb_q_vector *q_vector) +{ +u32 eics =0 ; + +if (!napi_if_scheduled_mark_missed (&q_vector->napi)) + eics = q_vector->eims_value; + +return eics; +} +int igb_xsk_wakeup (struct net_device *dev,u32 qid,u32 flags) {struct igb_adapter *adapter =netdev_priv (dev);
@@ -542,20 +552,32 @@ int igb_xsk_wakeup(struct net_device *dev, u32 qid, u32 flags) ring = adapter->tx_ring[qid]; -if (test_bit (IGB_RING_FLAG_TX_DISABLED , &ring->flags)) -return -ENETDOWN ; -if (!READ_ONCE (ring->xsk_pool))return -EINVAL ; -if (!napi_if_scheduled_mark_missed (&ring->q_vector->napi)) { +if (flags &XDP_WAKEUP_TX ) { +if (test_bit (IGB_RING_FLAG_TX_DISABLED , &ring->flags)) +return -ENETDOWN ; + + eics |=igb_sw_irq_prep (ring->q_vector); + } + +if (flags &XDP_WAKEUP_RX ) { +/* If IGB_FLAG_QUEUE_PAIRS is active, the q_vector +* and NAPI is shared between RX and TX. +* If NAPI is already running it would be marked as missed +* from the TX path, making this RX call a NOP +*/ + ring = adapter->rx_ring[qid]; + eics |=igb_sw_irq_prep (ring->q_vector); + } + +if (eics) {/* Cause software interrupt */ -if (adapter->flags &IGB_FLAG_HAS_MSIX ) { - eics |= ring->q_vector->eims_value; +if (adapter->flags &IGB_FLAG_HAS_MSIX )wr32 (E1000_EICS , eics); - }else { +else wr32 (E1000_ICS ,E1000_ICS_RXDMT0 ); - } }return 0 ;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 252/311 igc: Fix trigger of incorrect irq in igc_xsk_wakeup function
In reply to Sasha LevinFrom: Vivek Behera <[email protected]>
[ Upstream commit 554a1c34c11a057d01819ce9bb04653a8ffc8071 ]
This patch addresses the issue where the igc_xsk_wakeup function was triggering an incorrect IRQ for tx-0 when the i226 is configured with only 2 combined queues or in an environment with 2 active CPU cores. This prevented XDP Zero-copy send functionality in such split IRQ configurations.
The fix implements the correct logic for extracting q_vectors saved during rx and tx ring allocation and utilizes flags provided by the ndo_xsk_wakeup API to trigger the appropriate IRQ.
--- drivers/net/ethernet/intel/igc/igc_main.c | 34 ++++++++++++++++------- drivers/net/ethernet/intel/igc/igc_ptp.c | 3 +- 2 files changed, 26 insertions(+), 11 deletions(-)
drivers/net/ethernet/intel/igc/igc_main.c +24 -10
@@ -6908,28 +6908,29 @@ static int igc_xdp_xmit(struct net_device *dev, int num_frames,return nxmit; } -static void igc_trigger_rxtxq_interrupt (struct igc_adapter *adapter, -struct igc_q_vector *q_vector) +static u32 igc_sw_irq_prep (struct igc_q_vector *q_vector) { -struct igc_hw *hw = &adapter->hw;u32 eics =0 ; - eics |= q_vector->eims_value; -wr32 (IGC_EICS , eics); +if (!napi_if_scheduled_mark_missed (&q_vector->napi)) + eics = q_vector->eims_value; + +return eics; }int igc_xsk_wakeup (struct net_device *dev,u32 queue_id,u32 flags) {struct igc_adapter *adapter =netdev_priv (dev); -struct igc_q_vector *q_vector; +struct igc_hw *hw = &adapter->hw;struct igc_ring *ring; +u32 eics =0 ;if (test_bit (__IGC_DOWN, &adapter->state))return -ENETDOWN ;if (!igc_xdp_is_enabled (adapter))return -ENXIO ; - +/* Check if queue_id is valid. Tx and Rx queue numbers are always same */ if (queue_id >= adapter->num_rx_queues)return -EINVAL ;
@@ -6938,9 +6939,22 @@ int igc_xsk_wakeup(struct net_device *dev, u32 queue_id, u32 flags)if (!ring->xsk_pool)return -ENXIO ; - q_vector = adapter->q_vector[queue_id]; -if (!napi_if_scheduled_mark_missed (&q_vector->napi)) -igc_trigger_rxtxq_interrupt (adapter, q_vector); +if (flags &XDP_WAKEUP_RX ) + eics |=igc_sw_irq_prep (ring->q_vector); + +if (flags &XDP_WAKEUP_TX ) { +/* If IGC_FLAG_QUEUE_PAIRS is active, the q_vector +* and NAPI is shared between RX and TX. +* If NAPI is already running it would be marked as missed +* from the RX path, making this TX call a NOP +*/ + ring = adapter->tx_ring[queue_id]; + eics |=igc_sw_irq_prep (ring->q_vector); + } + +if (eics) +/* Cause software interrupt */ +wr32 (IGC_EICS , eics);return 0 ; }
drivers/net/ethernet/intel/igc/igc_ptp.c +2 -1
@@ -550,7 +550,8 @@ static void igc_ptp_free_tx_buffer(struct igc_adapter *adapter, tstamp->buffer_type =0 ;/* Trigger txrx interrupt for transmit completion */ -igc_xsk_wakeup (adapter->netdev, tstamp->xsk_queue_index,0 ); +igc_xsk_wakeup (adapter->netdev, tstamp->xsk_queue_index, +XDP_WAKEUP_TX );return ; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 253/311 bpf: Fix a UAF issue in bpf_trampoline_link_cgroup_shim
In reply to Sasha LevinFrom: Lang Xu <[email protected]>
[ Upstream commit 56145d237385ca0e7ca9ff7b226aaf2eb8ef368b ]
The root cause of this bug is that when 'bpf_link_put' reduces the refcount of 'shim_link->link.link' to zero, the resource is considered released but may still be referenced via 'tr->progs_hlist' in 'cgroup_shim_find'. The actual cleanup of 'tr->progs_hlist' in 'bpf_shim_tramp_link_release' is deferred. During this window, another process can cause a use-after-free via 'bpf_trampoline_link_cgroup_shim'.
Based on Martin KaFai Lau's suggestions, I have created a simple patch.
To fix this:
Add an atomic non-zero check in 'bpf_trampoline_link_cgroup_shim'. Only increment the refcount if it is not already zero.
Testing:
I verified the fix by adding a delay in 'bpf_shim_tramp_link_release' to make the bug easier to trigger:
static void bpf_shim_tramp_link_release(struct bpf_link *link)
{
/* ... */
if (!shim_link->trampoline)
return;+ msleep(100); WARN_ON_ONCE(bpf_trampoline_unlink_prog(&shim_link->link, shim_link->trampoline, NULL)); bpf_trampoline_put(shim_link->trampoline); }
Before the patch, running a PoC easily reproduced the crash(almost 100%) with a call trace similar to KaiyanM's report.
After the patch, the bug no longer occurs even after millions of iterations.
--- kernel/bpf/trampoline.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-)
kernel/bpf/trampoline.c +1 -3
@@ -796,10 +796,8 @@ int bpf_trampoline_link_cgroup_shim(struct bpf_prog *prog,mutex_lock (&tr->mutex); shim_link =cgroup_shim_find (tr, bpf_func); -if (shim_link) { +if (shim_link && !IS_ERR (bpf_link_inc_not_zero (&shim_link->link.link))) {/* Reusing existing shim attached by the other program. */ -bpf_link_inc (&shim_link->link.link); -mutex_unlock (&tr->mutex);bpf_trampoline_put (tr);/* bpf_trampoline_get above */ return 0 ;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 254/311 smb/client: fix buffer size for smb311_posix_qinfo in smb2_compound_op()
In reply to Sasha LevinFrom: ZhangGuoDong <[email protected]>
[ Upstream commit 12c43a062acb0ac137fc2a4a106d4d084b8c5416 ]
Use sizeof(struct smb311_posix_qinfo) instead of sizeof its pointer, so the allocated buffer matches the actual struct size.
--- fs/smb/client/smb2inode.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
fs/smb/client/smb2inode.c +2 -2
@@ -325,7 +325,7 @@ static int smb2_compound_op(const unsigned int xid, struct cifs_tcon *tcon, cfile->fid.volatile_fid,SMB_FIND_FILE_POSIX_INFO ,SMB2_O_INFO_FILE ,0 , -sizeof (struct smb311_posix_qinfo *) + +sizeof (struct smb311_posix_qinfo ) + (PATH_MAX *2 ) + (sizeof(struct smb_sid) * 2), 0, NULL); }else {
@@ -335,7 +335,7 @@ static int smb2_compound_op(const unsigned int xid, struct cifs_tcon *tcon,COMPOUND_FID ,SMB_FIND_FILE_POSIX_INFO ,SMB2_O_INFO_FILE ,0 , -sizeof (struct smb311_posix_qinfo *) + +sizeof (struct smb311_posix_qinfo ) + (PATH_MAX *2 ) + (sizeof(struct smb_sid) * 2), 0, NULL); }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 255/311 smb/client: fix buffer size for smb311_posix_qinfo in SMB311_posix_query_info()
In reply to Sasha LevinFrom: ZhangGuoDong <[email protected]>
[ Upstream commit 9621b996e4db1dbc2b3dc5d5910b7d6179397320 ]
SMB311_posix_query_info() is currently unused, but it may still be used in some stable versions, so these changes are submitted as a separate patch.
Use sizeof(struct smb311_posix_qinfo) instead of sizeof its pointer, so the allocated buffer matches the actual struct size.
--- fs/smb/client/smb2pdu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
fs/smb/client/smb2pdu.c +1 -1
@@ -3981,7 +3981,7 @@ SMB311_posix_query_info(const unsigned int xid, struct cifs_tcon *tcon, u64 persistent_fid, u64 volatile_fid, struct smb311_posix_qinfo *data, u32 *plen) { - size_t output_len = sizeof(struct smb311_posix_qinfo *) + + size_t output_len = sizeof(struct smb311_posix_qinfo) + (sizeof (struct smb_sid ) *2 ) + (PATH_MAX *2 ); *plen =0 ;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 256/311 ipv6: fix NULL pointer deref in ip6_rt_get_dev_rcu()
In reply to Sasha LevinFrom: Jakub Kicinski <[email protected]>
[ Upstream commit 2ffb4f5c2ccb2fa1c049dd11899aee7967deef5a ]
l3mdev_master_dev_rcu() can return NULL when the slave device is being un-slaved from a VRF. All other callers deal with this, but we lost the fallback to loopback in ip6_rt_pcpu_alloc() -> ip6_rt_get_dev_rcu() with commit 4832c30d5458 ("net: ipv6: put host and anycast routes on device with address").
KASAN: null-ptr-deref in range [0x0000000000000108-0x000000000000010f] RIP: 0010:ip6_rt_pcpu_alloc (net/ipv6/route.c:1418) Call Trace: ip6_pol_route (net/ipv6/route.c:2318) fib6_rule_lookup (net/ipv6/fib6_rules.c:115) ip6_route_output_flags (net/ipv6/route.c:2607) vrf_process_v6_outbound (drivers/net/vrf.c:437)
I was tempted to rework the un-slaving code to clear the flag first and insert synchronize_rcu() before we remove the upper. But looks like the explicit fallback to loopback_dev is an established pattern. And I guess avoiding the synchronize_rcu() is nice, too.
--- net/ipv6/route.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
net/ipv6/route.c +2 -1
@@ -1063,7 +1063,8 @@ static struct net_device *ip6_rt_get_dev_rcu(const struct fib6_result *res)*/ if (netif_is_l3_slave (dev) && !rt6_need_strict (&res->f6i->fib6_dst.addr)) - dev =l3mdev_master_dev_rcu (dev); + dev =l3mdev_master_dev_rcu (dev) ? : +dev_net (dev)->loopback_dev;else if (!netif_is_l3_master (dev)) dev =dev_net (dev)->loopback_dev;/* last case is netif_is_l3_master(dev) is true in which
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 257/311 net: ipv4: fix ARM64 alignment fault in multipath hash seed
In reply to Sasha LevinFrom: Yung Chih Su <[email protected]>
[ Upstream commit 4ee7fa6cf78ff26d783d39e2949d14c4c1cd5e7f ]
struct sysctl_fib_multipath_hash_seed contains two u32 fields (user_seed and mp_seed), making it an 8-byte structure with a 4-byte alignment requirement.
In fib_multipath_hash_from_keys(), the code evaluates the entire struct atomically via READ_ONCE():
mp_seed = READ_ONCE(net->ipv4.sysctl_fib_multipath_hash_seed).mp_seed;
While this silently works on GCC by falling back to unaligned regular loads which the ARM64 kernel tolerates, it causes a fatal kernel panic when compiled with Clang and LTO enabled.
Commit e35123d83ee3 ("arm64: lto: Strengthen READ_ONCE() to acquire when CONFIG_LTO=y") strengthens READ_ONCE() to use Load-Acquire instructions (ldar / ldapr) to prevent compiler reordering bugs under Clang LTO. Since the macro evaluates the full 8-byte struct, Clang emits a 64-bit ldar instruction. ARM64 architecture strictly requires ldar to be naturally aligned, thus executing it on a 4-byte aligned address triggers a strict Alignment Fault (FSC = 0x21).
Fix the read side by moving the READ_ONCE() directly to the u32 member, which emits a safe 32-bit ldar Wn.
Furthermore, Eric Dumazet pointed out that WRITE_ONCE() on the entire struct in proc_fib_multipath_hash_set_seed() is also flawed. Analysis shows that Clang splits this 8-byte write into two separate 32-bit str instructions. While this avoids an alignment fault, it destroys atomicity and exposes a tear-write vulnerability. Fix this by explicitly splitting the write into two 32-bit WRITE_ONCE() operations.
Finally, add the missing READ_ONCE() when reading user_seed in proc_fib_multipath_hash_seed() to ensure proper pairing and concurrency safety.
--- include/net/ip_fib.h | 2 +- net/ipv4/sysctl_net_ipv4.c | 5 +++-- 2 files changed, 4 insertions(+), 3 deletions(-)
include/net/ip_fib.h +1 -1
@@ -559,7 +559,7 @@ static inline u32 fib_multipath_hash_from_keys(const struct net *net,siphash_aligned_key_t hash_key;u32 mp_seed; - mp_seed =READ_ONCE (net->ipv4.sysctl_fib_multipath_hash_seed).mp_seed; + mp_seed =READ_ONCE (net->ipv4.sysctl_fib_multipath_hash_seed.mp_seed);fib_multipath_hash_construct_key (&hash_key, mp_seed);return flow_hash_from_keys_seed (keys, &hash_key);
net/ipv4/sysctl_net_ipv4.c +3 -2
@@ -486,7 +486,8 @@ static void proc_fib_multipath_hash_set_seed(struct net *net, u32 user_seed) proc_fib_multipath_hash_rand_seed), }; -WRITE_ONCE (net->ipv4.sysctl_fib_multipath_hash_seed, new); +WRITE_ONCE (net->ipv4.sysctl_fib_multipath_hash_seed.user_seed, new.user_seed); +WRITE_ONCE (net->ipv4.sysctl_fib_multipath_hash_seed.mp_seed, new.mp_seed); } static int proc_fib_multipath_hash_seed(const struct ctl_table *table, int write,
@@ -500,7 +501,7 @@ static int proc_fib_multipath_hash_seed(const struct ctl_table *table, int writeint ret; mphs = &net->ipv4.sysctl_fib_multipath_hash_seed; - user_seed = mphs->user_seed; + user_seed =READ_ONCE (mphs->user_seed); tmp = *table; tmp.data = &user_seed;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 258/311 amd-xgbe: fix sleep while atomic on suspend/resume
In reply to Sasha LevinFrom: Raju Rangoju <[email protected]>
[ Upstream commit e2f27363aa6d983504c6836dd0975535e2e9dba0 ]
The xgbe_powerdown() and xgbe_powerup() functions use spinlocks (spin_lock_irqsave) while calling functions that may sleep:
- napi_disable() can sleep waiting for NAPI polling to complete
- flush_workqueue() can sleep waiting for pending work items
This causes a "BUG: scheduling while atomic" error during suspend/resume cycles on systems using the AMD XGBE Ethernet controller.
The spinlock protection in these functions is unnecessary as these functions are called from suspend/resume paths which are already serialized by the PM core
Fix this by removing the spinlock. Since only code that takes this lock is xgbe_powerdown() and xgbe_powerup(), remove it completely.
--- drivers/net/ethernet/amd/xgbe/xgbe-drv.c | 10 ---------- drivers/net/ethernet/amd/xgbe/xgbe-main.c | 1 - drivers/net/ethernet/amd/xgbe/xgbe.h | 3 --- 3 files changed, 14 deletions(-)
drivers/net/ethernet/amd/xgbe/xgbe-drv.c +0 -10
@@ -1120,7 +1120,6 @@ int xgbe_powerdown(struct net_device *netdev, unsigned int caller) {struct xgbe_prv_data *pdata =netdev_priv (netdev);struct xgbe_hw_if *hw_if = &pdata->hw_if; -unsigned long flags;DBGPR ("-->xgbe_powerdown\n" );
@@ -1131,8 +1130,6 @@ int xgbe_powerdown(struct net_device *netdev, unsigned int caller)return -EINVAL ; } -spin_lock_irqsave (&pdata->lock, flags); -if (caller ==XGMAC_DRIVER_CONTEXT )netif_device_detach (netdev);
@@ -1148,8 +1145,6 @@ int xgbe_powerdown(struct net_device *netdev, unsigned int caller) pdata->power_down =1 ; -spin_unlock_irqrestore (&pdata->lock, flags); -DBGPR ("<--xgbe_powerdown\n" );return 0 ;
@@ -1159,7 +1154,6 @@ int xgbe_powerup(struct net_device *netdev, unsigned int caller) {struct xgbe_prv_data *pdata =netdev_priv (netdev);struct xgbe_hw_if *hw_if = &pdata->hw_if; -unsigned long flags;DBGPR ("-->xgbe_powerup\n" );
@@ -1170,8 +1164,6 @@ int xgbe_powerup(struct net_device *netdev, unsigned int caller)return -EINVAL ; } -spin_lock_irqsave (&pdata->lock, flags); - pdata->power_down =0 ;xgbe_napi_enable (pdata,0 );
@@ -1186,8 +1178,6 @@ int xgbe_powerup(struct net_device *netdev, unsigned int caller)xgbe_start_timers (pdata); -spin_unlock_irqrestore (&pdata->lock, flags); -DBGPR ("<--xgbe_powerup\n" );return 0 ;
drivers/net/ethernet/amd/xgbe/xgbe-main.c +0 -1
@@ -76,7 +76,6 @@ struct xgbe_prv_data *xgbe_alloc_pdata(struct device *dev) pdata->netdev = netdev; pdata->dev = dev; -spin_lock_init (&pdata->lock);spin_lock_init (&pdata->xpcs_lock);mutex_init (&pdata->rss_mutex);spin_lock_init (&pdata->tstamp_lock);
drivers/net/ethernet/amd/xgbe/xgbe.h +0 -3
@@ -1003,9 +1003,6 @@ struct xgbe_prv_data {unsigned int pp3;unsigned int pp4; -/* Overall device lock */ -spinlock_t lock; -/* XPCS indirect addressing lock */ spinlock_t xpcs_lock;unsigned int xpcs_window_def_reg;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 259/311 drm/sched: Fix kernel-doc warning for drm_sched_job_done()
In reply to Sasha LevinFrom: Yujie Liu <[email protected]>
[ Upstream commit 61ded1083b264ff67ca8c2de822c66b6febaf9a8 ]
There is a kernel-doc warning for the scheduler:
Warning: drivers/gpu/drm/scheduler/sched_main.c:367 function parameter 'result' not described in 'drm_sched_job_done'
Fix the warning by describing the undocumented error code.
--- drivers/gpu/drm/scheduler/sched_main.c | 1 + 1 file changed, 1 insertion(+)
drivers/gpu/drm/scheduler/sched_main.c +1 -0
@@ -361,6 +361,7 @@ static void drm_sched_run_free_queue(struct drm_gpu_scheduler *sched)/** * drm_sched_job_done - complete a job * @s_job: pointer to the job which is done +* @result: 0 on success, -ERRNO on error * * Finish the job's fence and resubmit the work items. */
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 260/311 ata: libata: cancel pending work after clearing deferred_qc
In reply to Sasha LevinFrom: Niklas Cassel <[email protected]>
[ Upstream commit aac9b27f7c1f2b2cf7f50a9ca633ecbbcaf22af9 ]
Syzbot reported a WARN_ON() in ata_scsi_deferred_qc_work(), caused by ap->ops->qc_defer() returning non-zero before issuing the deferred qc.
ata_scsi_schedule_deferred_qc() is called during each command completion. This function will check if there is a deferred QC, and if ap->ops->qc_defer() returns zero, meaning that it is possible to queue the deferred qc at this time (without being deferred), then it will queue the work which will issue the deferred qc.
Once the work get to run, which can potentially be a very long time after the work was scheduled, there is a WARN_ON() if ap->ops->qc_defer() returns non-zero.
While we hold the ap->lock both when assigning and clearing deferred_qc, and the work itself holds the ap->lock, the code currently does not cancel the work after clearing the deferred qc.
This means that the following scenario can happen:
1) One or several NCQ commands are queued.
2) A non-NCQ command is queued, gets stored in ap->deferred_qc.
3) Last NCQ command gets completed, work is queued to issue the deferred qc.
4) Timeout or error happens, ap->deferred_qc is cleared. The queued work is currently NOT canceled.
5) Port is reset.
6) One or several NCQ commands are queued.
7) A non-NCQ command is queued, gets stored in ap->deferred_qc.
8) Work is finally run. Yet at this time, there is still NCQ commands in flight.
The work in 8) really belongs to the non-NCQ command in 2), not to the non-NCQ command in 7). The reason why the work is executed when it is not supposed to, is because it was never canceled when ap->deferred_qc was cleared in 4). Thus, ensure that we always cancel the work after clearing ap->deferred_qc.
Another potential fix would have been to let ata_scsi_deferred_qc_work() do nothing if ap->ops->qc_defer() returns non-zero. However, canceling the work when clearing ap->deferred_qc seems slightly more logical, as we hold the ap->lock when clearing ap->deferred_qc, so we know that the work cannot be holding the lock. (The function could be waiting for the lock, but that is okay since it will do nothing if ap->deferred_qc is not set.)
--- drivers/ata/libata-eh.c | 1 + drivers/ata/libata-scsi.c | 1 + 2 files changed, 2 insertions(+)
drivers/ata/libata-eh.c +1 -0
@@ -659,6 +659,7 @@ void ata_scsi_cmd_error_handler(struct Scsi_Host *host, struct ata_port *ap,*/ WARN_ON_ONCE (qc->flags &ATA_QCFLAG_ACTIVE ); ap->deferred_qc =NULL ; +cancel_work (&ap->deferred_qc_work);set_host_byte (scmd,DID_TIME_OUT );scsi_eh_finish_cmd (scmd, &ap->eh_done_q); }else if (i <ATA_MAX_QUEUE ) {
drivers/ata/libata-scsi.c +1 -0
@@ -1699,6 +1699,7 @@ void ata_scsi_requeue_deferred_qc(struct ata_port *ap) scmd = qc->scsicmd; ap->deferred_qc =NULL ; +cancel_work (&ap->deferred_qc_work);ata_qc_free (qc); scmd->result = (DID_SOFT_ERROR <<16 );scsi_done (scmd);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 261/311 i2c: i801: Revert "i2c: i801: replace acpi_lock with I2C bus lock"
In reply to Sasha LevinFrom: Charles Haithcock <[email protected]>
[ Upstream commit cfc69c2e6c699c96949f7b0455195b0bfb7dc715 ]
This reverts commit f707d6b9e7c18f669adfdb443906d46cfbaaa0c1.
Under rare circumstances, multiple udev threads can collect i801 device info on boot and walk i801_acpi_io_handler somewhat concurrently. The first will note the area is reserved by acpi to prevent further touches. This ultimately causes the area to be deregistered. The second will enter i801_acpi_io_handler after the area is unregistered but before a check can be made that the area is unregistered. i2c_lock_bus relies on the now unregistered area containing lock_ops to lock the bus. The end result is a kernel panic on boot with the following backtrace;
[ 14.971872] ioatdma 0000:09:00.2: enabling device (0100 -> 0102) [ 14.971873] BUG: kernel NULL pointer dereference, address: 0000000000000000 [ 14.971880] #PF: supervisor read access in kernel mode [ 14.971884] #PF: error_code(0x0000) - not-present page [ 14.971887] PGD 0 P4D 0 [ 14.971894] Oops: 0000 [#1] PREEMPT SMP PTI [ 14.971900] CPU: 5 PID: 956 Comm: systemd-udevd Not tainted 5.14.0-611.5.1.el9_7.x86_64 #1 [ 14.971905] Hardware name: XXXXXXXXXXXXXXXXXXXXXXX BIOS 1.20.10.SV91 01/30/2023 [ 14.971908] RIP: 0010:i801_acpi_io_handler+0x2d/0xb0 [i2c_i801] [ 14.971929] Code: 00 00 49 8b 40 20 41 57 41 56 4d 8b b8 30 04 00 00 49 89 ce 41 55 41 89 d5 41 54 49 89 f4 be 02 00 00 00 55 4c 89 c5 53 89 fb <48> 8b 00 4c 89 c7 e8 18 61 54 e9 80 bd 80 04 00 00 00 75 09 4c 3b [ 14.971933] RSP: 0018:ffffbaa841483838 EFLAGS: 00010282 [ 14.971938] RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffff9685e01ba568 [ 14.971941] RDX: 0000000000000008 RSI: 0000000000000002 RDI: 0000000000000000 [ 14.971944] RBP: ffff9685ca22f028 R08: ffff9685ca22f028 R09: ffff9685ca22f028 [ 14.971948] R10: 000000000000000b R11: 0000000000000580 R12: 0000000000000580 [ 14.971951] R13: 0000000000000008 R14: ffff9685e01ba568 R15: ffff9685c222f000 [ 14.971954] FS: 00007f8287c0ab40(0000) GS:ffff96a47f940000(0000) knlGS:0000000000000000 [ 14.971959] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 14.971963] CR2: 0000000000000000 CR3: 0000000168090001 CR4: 00000000003706f0 [ 14.971966] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 14.971968] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 [ 14.971972] Call Trace: [ 14.971977] <TASK> [ 14.971981] ? show_trace_log_lvl+0x1c4/0x2df [ 14.971994] ? show_trace_log_lvl+0x1c4/0x2df [ 14.972003] ? acpi_ev_address_space_dispatch+0x16e/0x3c0 [ 14.972014] ? __die_body.cold+0x8/0xd [ 14.972021] ? page_fault_oops+0x132/0x170 [ 14.972028] ? exc_page_fault+0x61/0x150 [ 14.972036] ? asm_exc_page_fault+0x22/0x30 [ 14.972045] ? i801_acpi_io_handler+0x2d/0xb0 [i2c_i801] [ 14.972061] acpi_ev_address_space_dispatch+0x16e/0x3c0 [ 14.972069] ? __pfx_i801_acpi_io_handler+0x10/0x10 [i2c_i801] [ 14.972085] acpi_ex_access_region+0x5b/0xd0 [ 14.972093] acpi_ex_field_datum_io+0x73/0x2e0 [ 14.972100] acpi_ex_read_data_from_field+0x8e/0x230 [ 14.972106] acpi_ex_resolve_node_to_value+0x23d/0x310 [ 14.972114] acpi_ds_evaluate_name_path+0xad/0x110 [ 14.972121] acpi_ds_exec_end_op+0x321/0x510 [ 14.972127] acpi_ps_parse_loop+0xf7/0x680 [ 14.972136] acpi_ps_parse_aml+0x17a/0x3d0 [ 14.972143] acpi_ps_execute_method+0x137/0x270 [ 14.972150] acpi_ns_evaluate+0x1f4/0x2e0 [ 14.972158] acpi_evaluate_object+0x134/0x2f0 [ 14.972164] acpi_evaluate_integer+0x50/0xe0 [ 14.972173] ? vsnprintf+0x24b/0x570 [ 14.972181] acpi_ac_get_state.part.0+0x23/0x70 [ 14.972189] get_ac_property+0x4e/0x60 [ 14.972195] power_supply_show_property+0x90/0x1f0 [ 14.972205] add_prop_uevent+0x29/0x90 [ 14.972213] power_supply_uevent+0x109/0x1d0 [ 14.972222] dev_uevent+0x10e/0x2f0 [ 14.972228] uevent_show+0x8e/0x100 [ 14.972236] dev_attr_show+0x19/0x40 [ 14.972246] sysfs_kf_seq_show+0x9b/0x100 [ 14.972253] seq_read_iter+0x120/0x4b0 [ 14.972262] ? selinux_file_permission+0x106/0x150 [ 14.972273] vfs_read+0x24f/0x3a0 [ 14.972284] ksys_read+0x5f/0xe0 [ 14.972291] do_syscall_64+0x5f/0xe0
...
The kernel panic is mitigated by setting limiting the count of udev children to 1. Revert to using the acpi_lock to continue protecting marking the area as owned by firmware without relying on a lock in a potentially unmapped region of memory.
--- drivers/i2c/busses/i2c-i801.c | 14 ++++++++++---- 1 file changed, 10 insertions(+), 4 deletions(-)
drivers/i2c/busses/i2c-i801.c +10 -4
@@ -310,9 +310,10 @@ struct i801_priv {/* * If set to true the host controller registers are reserved for -* ACPI AML use. +* ACPI AML use. Needs extra protection by acpi_lock. */ bool acpi_reserved; +struct mutex acpi_lock; };#define FEATURE_SMBUS_PEC BIT(0)
@@ -894,8 +895,11 @@ static s32 i801_access(struct i2c_adapter *adap, u16 addr,int hwpec, ret;struct i801_priv *priv =i2c_get_adapdata (adap); -if (priv->acpi_reserved) +mutex_lock (&priv->acpi_lock); +if (priv->acpi_reserved) { +mutex_unlock (&priv->acpi_lock);return -EBUSY ; + }pm_runtime_get_sync (&priv->pci_dev->dev);
@@ -935,6 +939,7 @@ static s32 i801_access(struct i2c_adapter *adap, u16 addr,iowrite8 (SMBHSTSTS_INUSE_STS |STATUS_FLAGS ,SMBHSTSTS (priv));pm_runtime_put_autosuspend (&priv->pci_dev->dev); +mutex_unlock (&priv->acpi_lock);return ret; }
@@ -1465,7 +1470,7 @@ i801_acpi_io_handler(u32 function, acpi_physical_address address, u32 bits,* further access from the driver itself. This device is now owned * by the system firmware. */ -i2c_lock_bus (&priv->adapter,I2C_LOCK_SEGMENT ); +mutex_lock (&priv->acpi_lock);if (!priv->acpi_reserved &&i801_acpi_is_smbus_ioport (priv, address)) { priv->acpi_reserved = true;
@@ -1485,7 +1490,7 @@ i801_acpi_io_handler(u32 function, acpi_physical_address address, u32 bits,else status =acpi_os_write_port (address, (u32)*value, bits); -i2c_unlock_bus (&priv->adapter,I2C_LOCK_SEGMENT ); +mutex_unlock (&priv->acpi_lock);return status; }
@@ -1545,6 +1550,7 @@ static int i801_probe(struct pci_dev *dev, const struct pci_device_id *id) priv->adapter.dev.parent = &dev->dev;acpi_use_parent_companion (&priv->adapter.dev); priv->adapter.retries =3 ; +mutex_init (&priv->acpi_lock); priv->pci_dev = dev; priv->features = id->driver_data;
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 262/311 time/jiffies: Fix sysctl file error on configurations where USER_HZ < HZ
In reply to Sasha LevinFrom: Gerd Rausch <[email protected]>
[ Upstream commit 6932256d3a3764f3a5e06e2cb8603be45b6a9fef ]
Commit 2dc164a48e6fd ("sysctl: Create converter functions with two new macros") incorrectly returns error to user space when jiffies sysctl converter is used. The old overflow check got replaced with an unconditional one:
+ if (USER_HZ < HZ)
+ return -EINVAL;
which will always be true on configurations with "USER_HZ < HZ".
Remove the check; it is no longer needed as clock_t_to_jiffies() returns ULONG_MAX for the overflow case and proc_int_u2k_conv_uop() checks for "> INT_MAX" after conversion
--- kernel/time/jiffies.c | 2 -- 1 file changed, 2 deletions(-)
kernel/time/jiffies.c +0 -2
@@ -162,8 +162,6 @@ EXPORT_SYMBOL(proc_dointvec_jiffies);int proc_dointvec_userhz_jiffies (const struct ctl_table *table,int dir,void *buffer,size_t *lenp,loff_t *ppos) { -if (SYSCTL_USER_TO_KERN (dir) &&USER_HZ <HZ ) -return -EINVAL ;return proc_dointvec_conv (table, dir, buffer, lenp, ppos, do_proc_int_conv_userhz_jiffies); }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 263/311 drm/xe/gsc: Fix GSC proxy cleanup on early initialization failure
In reply to Sasha LevinFrom: Zhanjun Dong <[email protected]>
[ Upstream commit b3368ecca9538b88ddf982ea99064860fd5add97 ]
xe_gsc_proxy_remove undoes what is done in both xe_gsc_proxy_init and xe_gsc_proxy_start; however, if we fail between those 2 calls, it is possible that the HW forcewake access hasn't been initialized yet and so we hit errors when the cleanup code tries to write GSC register. To avoid that, split the cleanup in 2 functions so that the HW cleanup is only called if the HW setup was completed successfully.
Since the HW cleanup (interrupt disabling) is now removed from xe_gsc_proxy_remove, the cleanup on error paths in xe_gsc_proxy_start must be updated to disable interrupts before returning.
--- drivers/gpu/drm/xe/xe_gsc_proxy.c | 43 +++++++++++++++++++++++++------ drivers/gpu/drm/xe/xe_gsc_types.h | 2 ++ 2 files changed, 37 insertions(+), 8 deletions(-)
drivers/gpu/drm/xe/xe_gsc_proxy.c +35 -8
@@ -435,16 +435,12 @@ static int proxy_channel_alloc(struct xe_gsc *gsc)return 0 ; } -static void xe_gsc_proxy_remove (void *arg) +static void xe_gsc_proxy_stop (struct xe_gsc *gsc) { -struct xe_gsc *gsc = arg;struct xe_gt *gt =gsc_to_gt (gsc);struct xe_device *xe =gt_to_xe (gt);unsigned int fw_ref =0 ; -if (!gsc->proxy.component_added) -return ; -/* disable HECI2 IRQs */ xe_pm_runtime_get (xe); fw_ref =xe_force_wake_get (gt_to_fw (gt),XE_FW_GSC );
@@ -458,6 +454,30 @@ static void xe_gsc_proxy_remove(void *arg)xe_pm_runtime_put (xe);xe_gsc_wait_for_worker_completion (gsc); + gsc->proxy.started = false; +} + +static void xe_gsc_proxy_remove (void *arg) +{ +struct xe_gsc *gsc = arg; +struct xe_gt *gt =gsc_to_gt (gsc); +struct xe_device *xe =gt_to_xe (gt); + +if (!gsc->proxy.component_added) +return ; + +/* +* GSC proxy start is an async process that can be ongoing during +* Xe module load/unload. Using devm managed action to register +* xe_gsc_proxy_stop could cause issues if Xe module unload has +* already started when the action is registered, potentially leading +* to the cleanup being called at the wrong time. Therefore, instead +* of registering a separate devm action to undo what is done in +* proxy start, we call it from here, but only if the start has +* completed successfully (tracked with the 'started' flag). +*/ +if (gsc->proxy.started) +xe_gsc_proxy_stop (gsc);component_del (xe->drm.dev, &xe_gsc_proxy_component_ops); gsc->proxy.component_added = false;
@@ -513,6 +533,7 @@ int xe_gsc_proxy_init(struct xe_gsc *gsc)*/ int xe_gsc_proxy_start (struct xe_gsc *gsc) { +struct xe_gt *gt =gsc_to_gt (gsc);int err;/* enable the proxy interrupt in the GSC shim layer */
@@ -524,12 +545,18 @@ int xe_gsc_proxy_start(struct xe_gsc *gsc)*/ err =xe_gsc_proxy_request_handler (gsc);if (err) -return err; + goto err_irq_disable;if (!xe_gsc_proxy_init_done (gsc)) { -xe_gt_err (gsc_to_gt (gsc),"GSC FW reports proxy init not completed\n" ); -return -EIO ; +xe_gt_err (gt,"GSC FW reports proxy init not completed\n" ); + err = -EIO ; + goto err_irq_disable; } + gsc->proxy.started = true;return 0 ; + +err_irq_disable: +gsc_proxy_irq_toggle (gsc, false); +return err; }
drivers/gpu/drm/xe/xe_gsc_types.h +2 -0
@@ -58,6 +58,8 @@ struct xe_gsc {struct mutex mutex;/** @proxy.component_added: whether the component has been added */ bool component_added; +/** @proxy.started: whether the proxy has been started */ +bool started;/** @proxy.bo: object to store message to and from the GSC */ struct xe_bo *bo;/** @proxy.to_gsc: map of the memory used to send messages to the GSC */
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 264/311 drm/xe/reg_sr: Fix leak on xa_store failure
In reply to Sasha LevinFrom: Shuicheng Lin <[email protected]>
[ Upstream commit 3091723785def05ebfe6a50866f87a044ae314ba ]
Free the newly allocated entry when xa_store() fails to avoid a memory leak on the error path.
v2: use goto fail_free. (Bala)
--- drivers/gpu/drm/xe/xe_reg_sr.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/gpu/drm/xe/xe_reg_sr.c +3 -1
@@ -101,10 +101,12 @@ int xe_reg_sr_add(struct xe_reg_sr *sr, *pentry = *e; ret =xa_err (xa_store (&sr->xa, idx, pentry,GFP_KERNEL ));if (ret) - goto fail; + goto fail_free;return 0 ; +fail_free: +kfree (pentry); fail: xe_gt_err(gt, "discarding save-restore reg %04lx (clear: %08x, set: %08x, masked: %s, mcr: %s): ret=%d\n",
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 265/311 nvme: fix memory allocation in nvme_pr_read_keys()
In reply to Sasha LevinFrom: Sungwoo Kim <[email protected]>
[ Upstream commit c3320153769f05fd7fe9d840cb555dd3080ae424 ]
nvme_pr_read_keys() takes num_keys from userspace and uses it to calculate the allocation size for rse via struct_size(). The upper limit is PR_KEYS_MAX (64K).
A malicious or buggy userspace can pass a large num_keys value that results in a 4MB allocation attempt at most, causing a warning in the page allocator when the order exceeds MAX_PAGE_ORDER.
To fix this, use kvzalloc() instead of kzalloc().
This bug has the same reasoning and fix with the patch below:
https://lore.kernel.org/linux-block/[email protected]/
Warning log:
WARNING: mm/page_alloc.c:5216 at __alloc_frozen_pages_noprof+0x5aa/0x2300 mm/page_alloc.c:5216, CPU#1: syz-executor117/272 Modules linked in: CPU: 1 UID: 0 PID: 272 Comm: syz-executor117 Not tainted 6.19.0 #1 PREEMPT(voluntary) Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/2014 RIP: 0010:__alloc_frozen_pages_noprof+0x5aa/0x2300 mm/page_alloc.c:5216 Code: ff 83 bd a8 fe ff ff 0a 0f 86 69 fb ff ff 0f b6 1d f9 f9 c4 04 80 fb 01 0f 87 3b 76 30 ff 83 e3 01 75 09 c6 05 e4 f9 c4 04 01 <0f> 0b 48 c7 85 70 fe ff ff 00 00 00 00 e9 8f fd ff ff 31 c0 e9 0d RSP: 0018:ffffc90000fcf450 EFLAGS: 00010246 RAX: 0000000000000000 RBX: 0000000000000000 RCX: 1ffff920001f9ea0 RDX: 0000000000000000 RSI: 000000000000000b RDI: 0000000000040dc0 RBP: ffffc90000fcf648 R08: ffff88800b6c3380 R09: 0000000000000001 R10: ffffc90000fcf840 R11: ffff88807ffad280 R12: 0000000000000000 R13: 0000000000040dc0 R14: 0000000000000001 R15: ffffc90000fcf620 FS: 0000555565db33c0(0000) GS:ffff8880be26c000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 000000002000000c CR3: 0000000003b72000 CR4: 00000000000006f0 Call Trace: <TASK> alloc_pages_mpol+0x236/0x4d0 mm/mempolicy.c:2486 alloc_frozen_pages_noprof+0x149/0x180 mm/mempolicy.c:2557 ___kmalloc_large_node+0x10c/0x140 mm/slub.c:5598 __kmalloc_large_node_noprof+0x25/0xc0 mm/slub.c:5629 __do_kmalloc_node mm/slub.c:5645 [inline] __kmalloc_noprof+0x483/0x6f0 mm/slub.c:5669 kmalloc_noprof include/linux/slab.h:961 [inline] kzalloc_noprof include/linux/slab.h:1094 [inline] nvme_pr_read_keys+0x8f/0x4c0 drivers/nvme/host/pr.c:245 blkdev_pr_read_keys block/ioctl.c:456 [inline] blkdev_common_ioctl+0x1b71/0x29b0 block/ioctl.c:730 blkdev_ioctl+0x299/0x700 block/ioctl.c:786 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:597 [inline] __se_sys_ioctl fs/ioctl.c:583 [inline] __x64_sys_ioctl+0x1bf/0x220 fs/ioctl.c:583 x64_sys_call+0x1280/0x21b0 mnt/fuzznvme_1/fuzznvme/linux-build/v6.19/./arch/x86/include/generated/asm/syscalls_64.h:17 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline] do_syscall_64+0x71/0x330 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x76/0x7e RIP: 0033:0x7fb893d3108d Code: 28 c3 e8 46 1e 00 00 66 0f 1f 44 00 00 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007ffff61f2f38 EFLAGS: 00000246 ORIG_RAX: 0000000000000010 RAX: ffffffffffffffda RBX: 00007ffff61f3138 RCX: 00007fb893d3108d RDX: 0000000020000040 RSI: 00000000c01070ce RDI: 0000000000000003 RBP: 0000000000000001 R08: 0000000000000000 R09: 00007ffff61f3138 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000001 R13: 00007ffff61f3128 R14: 00007fb893dae530 R15: 0000000000000001 </TASK>
--- drivers/nvme/host/pr.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/nvme/host/pr.c +2 -2
@@ -242,7 +242,7 @@ static int nvme_pr_read_keys(struct block_device *bdev,if (rse_len >U32_MAX )return -EINVAL ; - rse =kzalloc (rse_len,GFP_KERNEL ); + rse =kvzalloc (rse_len,GFP_KERNEL );if (!rse)return -ENOMEM ;
@@ -267,7 +267,7 @@ static int nvme_pr_read_keys(struct block_device *bdev, } free_rse: -kfree (rse); +kvfree (rse);return ret; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 266/311 x86/numa: Store extra copy of numa_nodes_parsed
In reply to Sasha LevinFrom: Peter Zijlstra <[email protected]>
[ Upstream commit 48084cc153a5b0fbf0aa98d47670d3be0b9f64d5 ]
The topology setup code needs to know the total number of physical nodes enumerated in SRAT; however NUMA_EMU can cause the existing numa_nodes_parsed bitmap to be fictitious. Therefore, keep a copy of the bitmap specifically to retain the physical node count.
Stable-dep-of: 528d89a4707e ("x86/topo: Fix SNC topology mess")
--- arch/x86/include/asm/numa.h | 6 ++++++ arch/x86/mm/numa.c | 8 ++++++++ arch/x86/mm/srat.c | 2 ++ 3 files changed, 16 insertions(+)
arch/x86/include/asm/numa.h +6 -0
@@ -22,6 +22,7 @@ extern int numa_off;*/ extern s16 __apicid_to_node[MAX_LOCAL_APIC ];extern nodemask_t numa_nodes_parsed__initdata ; +extern nodemask_t numa_phys_nodes_parsed__initdata ;static inline void set_apicid_to_node (int apicid,s16 node) {
@@ -48,6 +49,7 @@ extern void __init init_cpu_to_node(void);extern void numa_add_cpu (unsigned int cpu);extern void numa_remove_cpu (unsigned int cpu);extern void init_gi_nodes (void ); +extern int num_phys_nodes (void );#else /* CONFIG_NUMA */ static inline void numa_set_node (int cpu,int node) { }static inline void numa_clear_node (int cpu) { }
@@ -55,6 +57,10 @@ static inline void init_cpu_to_node(void) { }static inline void numa_add_cpu (unsigned int cpu) { }static inline void numa_remove_cpu (unsigned int cpu) { }static inline void init_gi_nodes (void ) { } +static inline int num_phys_nodes (void ) +{ +return 1 ; +}#endif /* CONFIG_NUMA */ #ifdef CONFIG_DEBUG_PER_CPU_MAPS
arch/x86/mm/numa.c +8 -0
@@ -48,6 +48,8 @@ s16 __apicid_to_node[MAX_LOCAL_APIC] = { [0 ... MAX_LOCAL_APIC-1] = NUMA_NO_NODE }; +nodemask_t numa_phys_nodes_parsed__initdata ; +int numa_cpu_node (int cpu) {u32 apicid =early_per_cpu (x86_cpu_to_apicid, cpu);
@@ -57,6 +59,11 @@ int numa_cpu_node(int cpu)return NUMA_NO_NODE ; } +int __init num_phys_nodes (void ) +{ +return bitmap_weight (numa_phys_nodes_parsed.bits,MAX_NUMNODES ); +} +cpumask_var_t node_to_cpumask_map[MAX_NUMNODES ];EXPORT_SYMBOL (node_to_cpumask_map);
@@ -210,6 +217,7 @@ static int __init dummy_numa_init(void) 0LLU, PFN_PHYS(max_pfn) - 1);node_set (0 , numa_nodes_parsed); +node_set (0 , numa_phys_nodes_parsed);numa_add_memblk (0 ,0 ,PFN_PHYS (max_pfn));return 0 ;
arch/x86/mm/srat.c +2 -0
@@ -57,6 +57,7 @@ acpi_numa_x2apic_affinity_init(struct acpi_srat_x2apic_cpu_affinity *pa) }set_apicid_to_node (apic_id, node);node_set (node, numa_nodes_parsed); +node_set (node, numa_phys_nodes_parsed);pr_debug ("SRAT: PXM %u -> APIC 0x%04x -> Node %u\n" , pxm, apic_id, node); }
@@ -97,6 +98,7 @@ acpi_numa_processor_affinity_init(struct acpi_srat_cpu_affinity *pa)set_apicid_to_node (apic_id, node);node_set (node, numa_nodes_parsed); +node_set (node, numa_phys_nodes_parsed);pr_debug ("SRAT: PXM %u -> APIC 0x%02x -> Node %u\n" , pxm, apic_id, node); }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 267/311 x86/topo: Add topology_num_nodes_per_package()
In reply to Sasha LevinFrom: Peter Zijlstra <[email protected]>
[ Upstream commit ae6730ff42b3a13d94b405edeb5e40108b6d21b6 ]
Use the MADT and SRAT table data to compute __num_nodes_per_package.
Specifically, SRAT has already been parsed in x86_numa_init(), which is called before acpi_boot_init() which parses MADT. So both are available in topology_init_possible_cpus().
This number is useful to divinate the various Intel CoD/SNC and AMD NPS modes, since the platforms are failing to provide this otherwise.
Doing it this way is independent of the number of online CPUs and other such shenanigans.
Stable-dep-of: 528d89a4707e ("x86/topo: Fix SNC topology mess")
--- arch/x86/include/asm/topology.h | 6 ++++++ arch/x86/kernel/cpu/common.c | 3 +++ arch/x86/kernel/cpu/topology.c | 13 +++++++++++-- 3 files changed, 20 insertions(+), 2 deletions(-)
arch/x86/include/asm/topology.h +6 -0
@@ -155,6 +155,7 @@ extern unsigned int __max_logical_packages;extern unsigned int __max_threads_per_core;extern unsigned int __num_threads_per_package;extern unsigned int __num_cores_per_package; +extern unsigned int __num_nodes_per_package;const char *get_topology_cpu_type_name (struct cpuinfo_x86 *c);enum x86_topology_cpu_type get_topology_cpu_type (struct cpuinfo_x86 *c);
@@ -179,6 +180,11 @@ static inline unsigned int topology_num_threads_per_package(void)return __num_threads_per_package; } +static inline unsigned int topology_num_nodes_per_package (void ) +{ +return __num_nodes_per_package; +} +#ifdef CONFIG_X86_LOCAL_APIC int topology_get_logical_id (u32 apicid,enum x86_topology_domains at_level);#else
arch/x86/kernel/cpu/common.c +3 -0
@@ -95,6 +95,9 @@ EXPORT_SYMBOL(__max_dies_per_package);unsigned int __max_logical_packages__ro_after_init =1 ;EXPORT_SYMBOL (__max_logical_packages); +unsigned int __num_nodes_per_package__ro_after_init =1 ; +EXPORT_SYMBOL (__num_nodes_per_package); +unsigned int __num_cores_per_package__ro_after_init =1 ;EXPORT_SYMBOL (__num_cores_per_package);
arch/x86/kernel/cpu/topology.c +11 -2
@@ -31,6 +31,7 @@#include <asm/mpspec.h> #include <asm/msr.h> #include <asm/smp.h> +#include <asm/numa.h> #include "cpu.h"
@@ -492,11 +493,19 @@ void __init topology_init_possible_cpus(void)set_nr_cpu_ids (allowed); cnta =domain_weight (TOPO_PKG_DOMAIN ); - cntb =domain_weight (TOPO_DIE_DOMAIN ); __max_logical_packages = cnta; + +pr_info ("Max. logical packages: %3u\n" , __max_logical_packages); + + cntb =num_phys_nodes (); + __num_nodes_per_package =DIV_ROUND_UP (cntb, cnta); + +pr_info ("Max. logical nodes: %3u\n" , cntb); +pr_info ("Num. nodes per package:%3u\n" , __num_nodes_per_package); + + cntb =domain_weight (TOPO_DIE_DOMAIN ); __max_dies_per_package =1U << (get_count_order (cntb) -get_count_order (cnta)); -pr_info ("Max. logical packages: %3u\n" , cnta);pr_info ("Max. logical dies: %3u\n" , cntb);pr_info ("Max. dies per package: %3u\n" , __max_dies_per_package);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 268/311 x86/topo: Replace x86_has_numa_in_package
In reply to Sasha LevinFrom: Peter Zijlstra <[email protected]>
[ Upstream commit 717b64d58cff6fb97f97be07e382ed7641167a56 ]
.. with the brand spanking new topology_num_nodes_per_package().
Having the topology setup determine this value during MADT/SRAT parsing before SMP bringup avoids having to detect this situation when building the SMP topology masks.
Stable-dep-of: 528d89a4707e ("x86/topo: Fix SNC topology mess")
--- arch/x86/kernel/smpboot.c | 13 +++---------- 1 file changed, 3 insertions(+), 10 deletions(-)
arch/x86/kernel/smpboot.c +3 -10
@@ -468,13 +468,6 @@ static int x86_cluster_flags(void) }#endif -/* -* Set if a package/die has multiple NUMA nodes inside. -* AMD Magny-Cours, Intel Cluster-on-Die, and Intel -* Sub-NUMA Clustering have this. -*/ -static bool x86_has_numa_in_package; -static struct sched_domain_topology_level x86_topology[] = {SDTL_INIT (tl_smt_mask, cpu_smt_flags,SMT ), #ifdef CONFIG_SCHED_CLUSTER
@@ -496,7 +489,7 @@ static void __init build_sched_topology(void)* PKG domain since the NUMA domains will auto-magically create the * right spanning domains based on the SLIT. */ -if (x86_has_numa_in_package) { +if (topology_num_nodes_per_package () >1 ) {unsigned int pkgdom =ARRAY_SIZE (x86_topology) -2 ;memset (&x86_topology[pkgdom],0 ,sizeof (x86_topology[pkgdom]));
@@ -550,7 +543,7 @@ int arch_sched_node_distance(int from, int to)case INTEL_GRANITERAPIDS_X :case INTEL_ATOM_DARKMONT_X : -if (!x86_has_numa_in_package ||topology_max_packages () ==1 || +if (topology_max_packages () ==1 ||topology_num_nodes_per_package () ==1 || d <REMOTE_DISTANCE )return d;
@@ -606,7 +599,7 @@ void set_cpu_sibling_map(int cpu) o = &cpu_data (i);if (match_pkg (c, o) && !topology_same_node (c, o)) - x86_has_numa_in_package = true; +WARN_ON_ONCE (topology_num_nodes_per_package () ==1 );if ((i == cpu) || (has_smt &&match_smt (c, o)))link_mask (topology_sibling_cpumask, cpu, i);
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 269/311 x86/topo: Fix SNC topology mess
In reply to Sasha LevinFrom: Peter Zijlstra <[email protected]>
[ Upstream commit 528d89a4707e5bfd86e30823c45dbb66877df900 ]
Per 4d6dd05d07d0 ("sched/topology: Fix sched domain build error for GNR, CWF in SNC-3 mode"), the original crazy SNC-3 SLIT table was:
node distances:
node 0 1 2 3 4 5
0: 10 15 17 21 28 26
1: 15 10 15 23 26 23
2: 17 15 10 26 23 21
3: 21 28 26 10 15 17
4: 23 26 23 15 10 15
5: 26 23 21 17 15 10And per:
https://lore.kernel.org/lkml/[email protected]/
The suggestion was to average the off-trace clusters to restore sanity.
However, 4d6dd05d07d0 implements this under various assumptions:
- anything GNR/CWF with numa_in_package;
- there will never be more than 2 packages;
- the off-trace cluster will have distance >20
And then HPE shows up with a machine that matches the Vendor-Family-Model checks but looks like this:
Here's an 8 socket (2 chassis) HPE system with SNC enabled:
node 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0: 10 12 16 16 16 16 18 18 40 40 40 40 40 40 40 40 1: 12 10 16 16 16 16 18 18 40 40 40 40 40 40 40 40 2: 16 16 10 12 18 18 16 16 40 40 40 40 40 40 40 40 3: 16 16 12 10 18 18 16 16 40 40 40 40 40 40 40 40 4: 16 16 18 18 10 12 16 16 40 40 40 40 40 40 40 40 5: 16 16 18 18 12 10 16 16 40 40 40 40 40 40 40 40 6: 18 18 16 16 16 16 10 12 40 40 40 40 40 40 40 40 7: 18 18 16 16 16 16 12 10 40 40 40 40 40 40 40 40 8: 40 40 40 40 40 40 40 40 10 12 16 16 16 16 18 18 9: 40 40 40 40 40 40 40 40 12 10 16 16 16 16 18 18 10: 40 40 40 40 40 40 40 40 16 16 10 12 18 18 16 16 11: 40 40 40 40 40 40 40 40 16 16 12 10 18 18 16 16 12: 40 40 40 40 40 40 40 40 16 16 18 18 10 12 16 16 13: 40 40 40 40 40 40 40 40 16 16 18 18 12 10 16 16 14: 40 40 40 40 40 40 40 40 18 18 16 16 16 16 10 12 15: 40 40 40 40 40 40 40 40 18 18 16 16 16 16 12 10
10 = Same chassis and socket
12 = Same chassis and socket (SNC)
16 = Same chassis and adjacent socket
18 = Same chassis and non-adjacent socket
40 = Different chassis
Turns out, the 'max 2 packages' thing is only relevant to the SNC-3 parts, the smaller parts do 8 sockets (like usual). The above SLIT table is sane, but violates the previous assumptions and trips a WARN.
Now that the topology code has a sensible measure of nodes-per-package, we can use that to divinate the SNC mode at hand, and only fix up SNC-3 topologies.
There is a 'healthy' amount of paranoia code validating the assumptions on the SLIT table, a simple pr_err(FW_BUG) print on failure and a fallback to using the regular table. Lets see how long this lasts :-)
--- arch/x86/kernel/smpboot.c | 190 ++++++++++++++++++++++++++++---------- 1 file changed, 143 insertions(+), 47 deletions(-)
arch/x86/kernel/smpboot.c +143 -47
@@ -506,33 +506,149 @@ static void __init build_sched_topology(void) }#ifdef CONFIG_NUMA -static int sched_avg_remote_distance; -static int avg_remote_numa_distance (void ) +/* +* Test if the on-trace cluster at (N,N) is symmetric. +* Uses upper triangle iteration to avoid obvious duplicates. +*/ +static bool slit_cluster_symmetric (int N ) { -int i, j; -int distance, nr_remote, total_distance; - -if (sched_avg_remote_distance >0 ) -return sched_avg_remote_distance; - - nr_remote =0 ; - total_distance =0 ; -for_each_node_state (i,N_CPU ) { -for_each_node_state (j,N_CPU ) { - distance =node_distance (i, j); - -if (distance >=REMOTE_DISTANCE ) { - nr_remote++; - total_distance += distance; - } +int u =topology_num_nodes_per_package (); + +for (int k =0 ; k < u; k++) { +for (int l = k; l < u; l++) { +if (node_distance (N + k,N + l) != +node_distance (N + l,N + k)) +return false; } } -if (nr_remote) - sched_avg_remote_distance = total_distance / nr_remote; -else - sched_avg_remote_distance =REMOTE_DISTANCE ; -return sched_avg_remote_distance; +return true; +} + +/* +* Return the package-id of the cluster, or ~0 if indeterminate. +* Each node in the on-trace cluster should have the same package-id. +*/ +static u32 slit_cluster_package (int N ) +{ +int u =topology_num_nodes_per_package (); +u32 pkg_id = ~0 ; + +for (int n =0 ; n < u; n++) { +const struct cpumask *cpus =cpumask_of_node (N + n); +int cpu; + +for_each_cpu (cpu, cpus) { +u32 id =topology_logical_package_id (cpu); + +if (pkg_id == ~0 ) + pkg_id = id; +if (pkg_id != id) +return ~0 ; + } + } + +return pkg_id; +} + +/* +* Validate the SLIT table is of the form expected for SNC, specifically: +* +* - each on-trace cluster should be symmetric, +* - each on-trace cluster should have a unique package-id. +* +* If you NUMA_EMU on top of SNC, you get to keep the pieces. +*/ +static bool slit_validate (void ) +{ +int u =topology_num_nodes_per_package (); +u32 pkg_id, prev_pkg_id = ~0 ; + +for (int pkg =0 ; pkg <topology_max_packages (); pkg++) { +int n = pkg * u; + +/* +* Ensure the on-trace cluster is symmetric and each cluster +* has a different package id. +*/ +if (!slit_cluster_symmetric (n)) +return false; + pkg_id =slit_cluster_package (n); +if (pkg_id == ~0 ) +return false; +if (pkg && pkg_id == prev_pkg_id) +return false; + + prev_pkg_id = pkg_id; + } + +return true; +} + +/* +* Compute a sanitized SLIT table for SNC; notably SNC-3 can end up with +* asymmetric off-trace clusters, reflecting physical assymmetries. However +* this leads to 'unfortunate' sched_domain configurations. +* +* For example dual socket GNR with SNC-3: +* +* node distances: +* node 0 1 2 3 4 5 +* 0: 10 15 17 21 28 26 +* 1: 15 10 15 23 26 23 +* 2: 17 15 10 26 23 21 +* 3: 21 28 26 10 15 17 +* 4: 23 26 23 15 10 15 +* 5: 26 23 21 17 15 10 +* +* Fix things up by averaging out the off-trace clusters; resulting in: +* +* node 0 1 2 3 4 5 +* 0: 10 15 17 24 24 24 +* 1: 15 10 15 24 24 24 +* 2: 17 15 10 24 24 24 +* 3: 24 24 24 10 15 17 +* 4: 24 24 24 15 10 15 +* 5: 24 24 24 17 15 10 +*/ +static int slit_cluster_distance (int i,int j) +{ +static int slit_valid =-1 ; +int u =topology_num_nodes_per_package (); +long d =0 ; +int x, y; + +if (slit_valid <0 ) { + slit_valid =slit_validate (); +if (!slit_valid) +pr_err (FW_BUG "SLIT table doesn't have the expected form for SNC -- fixup disabled!\n" ); +else +pr_info ("Fixing up SNC SLIT table.\n" ); + } + +/* +* Is this a unit cluster on the trace? +*/ +if ((i / u) == (j / u) || !slit_valid) +return node_distance (i, j); + +/* +* Off-trace cluster. +* +* Notably average out the symmetric pair of off-trace clusters to +* ensure the resulting SLIT table is symmetric. +*/ + x = i - (i % u); + y = j - (j % u); + +for (i = x; i < x + u; i++) { +for (j = y; j < y + u; j++) { + d +=node_distance (i, j); + d +=node_distance (j, i); + } + } + +return d / (2 *u*u); }int arch_sched_node_distance (int from,int to)
@@ -542,34 +658,14 @@ int arch_sched_node_distance(int from, int to)switch (boot_cpu_data.x86_vfm) {case INTEL_GRANITERAPIDS_X :case INTEL_ATOM_DARKMONT_X : - -if (topology_max_packages () ==1 ||topology_num_nodes_per_package () ==1 || - d <REMOTE_DISTANCE ) +if (topology_max_packages () ==1 || +topology_num_nodes_per_package () <3 )return d;/* -* With SNC enabled, there could be too many levels of remote -* NUMA node distances, creating NUMA domain levels -* including local nodes and partial remote nodes. -* -* Trim finer distance tuning for NUMA nodes in remote package -* for the purpose of building sched domains. Group NUMA nodes -* in the remote package in the same sched group. -* Simplify NUMA domains and avoid extra NUMA levels including -* different remote NUMA nodes and local nodes. -* -* GNR and CWF don't expect systems with more than 2 packages -* and more than 2 hops between packages. Single average remote -* distance won't be appropriate if there are more than 2 -* packages as average distance to different remote packages -* could be different. +* Handle SNC-3 asymmetries. */ -WARN_ONCE (topology_max_packages () >2 , -"sched: Expect only up to 2 packages for GNR or CWF, " -"but saw %d packages when building sched domains." , -topology_max_packages ()); - - d =avg_remote_numa_distance (); +return slit_cluster_distance (from, to); }return d; }
Signature · 1 line
2.51.0
Sasha Levin
maintainerPATCH 270/311 sched/deadline: Fix missing ENQUEUE_REPLENISH during PI de-boosting
In reply to Sasha LevinFrom: Juri Lelli <[email protected]>
[ Upstream commit d658686a1331db3bb108ca079d76deb3208ed949 ]
Running stress-ng --schedpolicy 0 on an RT kernel on a big machine might lead to the following WARNINGs (edited).
sched: DL de-boosted task PID 22725: REPLENISH flag missing
WARNING: CPU: 93 PID: 0 at kernel/sched/deadline.c:239 dequeue_task_dl+0x15c/0x1f8 ... (running_bw underflow) Call trace: dequeue_task_dl+0x15c/0x1f8 (P) dequeue_task+0x80/0x168 deactivate_task+0x24/0x50 push_dl_task+0x264/0x2e0 dl_task_timer+0x1b0/0x228 __hrtimer_run_queues+0x188/0x378 hrtimer_interrupt+0xfc/0x260 ...
The problem is that when a SCHED_DEADLINE task (lock holder) is changed to a lower priority class via sched_setscheduler(), it may fail to properly inherit the parameters of potential DEADLINE donors if it didn't already inherit them in the past (shorter deadline than donor's at that time). This might lead to bandwidth accounting corruption, as enqueue_task_dl() won't recognize the lock holder as boosted.
The scenario occurs when:
1. A DEADLINE task (donor) blocks on a PI mutex held by another DEADLINE task (holder), but the holder doesn't inherit parameters (e.g., it already has a shorter deadline)
2. sched_setscheduler() changes the holder from DEADLINE to a lower class while still holding the mutex
3. The holder should now inherit DEADLINE parameters from the donor and be enqueued with ENQUEUE_REPLENISH, but this doesn't happen
Fix the issue by introducing __setscheduler_dl_pi(), which detects when a DEADLINE (proper or boosted) task gets setscheduled to a lower priority class. In case, the function makes the task inherit DEADLINE parameters of the donoer (pi_se) and sets ENQUEUE_REPLENISH flag to ensure proper bandwidth accounting during the next enqueue operation.
--- kernel/sched/syscalls.c | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+)
kernel/sched/syscalls.c +30 -0
@@ -284,6 +284,35 @@ static bool check_same_owner(struct task_struct *p) uid_eq(cred->euid, pcred->uid)); } +#ifdef CONFIG_RT_MUTEXES +static inline void __setscheduler_dl_pi (int newprio,int policy, +struct task_struct *p, +struct sched_change_ctx *scope) +{ +/* +* In case a DEADLINE task (either proper or boosted) gets +* setscheduled to a lower priority class, check if it neeeds to +* inherit parameters from a potential pi_task. In that case make +* sure replenishment happens with the next enqueue. +*/ + +if (dl_prio (newprio) && !dl_policy (policy)) { +struct task_struct *pi_task =rt_mutex_get_top_task (p); + +if (pi_task) { + p->dl.pi_se = pi_task->dl.pi_se; + scope->flags |=ENQUEUE_REPLENISH ; + } + } +} +#else /* !CONFIG_RT_MUTEXES */ +static inline void __setscheduler_dl_pi (int newprio,int policy, +struct task_struct *p, +struct sched_change_ctx *scope) +{ +} +#endif /* !CONFIG_RT_MUTEXES */ +#ifdef CONFIG_UCLAMP_TASK static int uclamp_validate(struct task_struct *p,
@@ -655,6 +684,7 @@ int __sched_setscheduler(struct task_struct *p,__setscheduler_params (p, attr); p->sched_class = next_class; p->prio = newprio; +__setscheduler_dl_pi (newprio, policy, p, scope); }__setscheduler_uclamp (p, attr);
Signature · 1 line
2.51.0