Skip to contentSkip to conversation

iommu/amd: Fix the configuration of GCR3 table root pointer

PATCH In review Sasha Levin 4 participants 84 messages Apr 18, 2020 – Apr 28, 2020

Subsystems (21)
lore
Share & tools
lore

Sasha Levin

author maintainer
#0 lore raw

PATCH 1/78 iommu/amd: Fix the configuration of GCR3 table root pointer

From: Adrian Huang <[email protected]>

[ Upstream commit c20f36534666e37858a14e591114d93cc1be0d34 ]

The SPA of the GCR3 table root pointer[51:31] masks 20 bits. However, this requires 21 bits (Please see the AMD IOMMU specification). This leads to the potential failure when the bit 51 of SPA of the GCR3 table root pointer is 1'.

Fixes 52815b75682e2 ("iommu/amd: Add support for IOMMUv2 domain mode")
---
 drivers/iommu/amd_iommu_types.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/iommu/amd_iommu_types.h +1 -1
index daeabd98c60e2..0679896b9e2e1 100644
@@ -348,7 +348,7 @@
 
 #define DTE_GCR3_VAL_A(x)	(((x) >> 12) & 0x00007ULL)
 #define DTE_GCR3_VAL_B(x)	(((x) >> 15) & 0x0ffffULL)
-#define DTE_GCR3_VAL_C(x)	(((x) >> 31) & 0xfffffULL)
+#define DTE_GCR3_VAL_C(x)	(((x) >> 31) & 0x1fffffULL)
 
 #define DTE_GCR3_INDEX_A	0
 #define DTE_GCR3_INDEX_B	1
Signature · 1 line
2.20.1
author maintainer
#1 lore raw

PATCH 2/78 tools/testing/nvdimm: Fix compilation failure without CONFIG_DEV_DAX_PMEM_COMPAT

From: Jan Kara <[email protected]>

[ Upstream commit c0e71d602053e4e7637e4bc7d0bc9603ea77a33f ]

When a kernel is configured without CONFIG_DEV_DAX_PMEM_COMPAT, the compilation of tools/testing/nvdimm fails with:

Building modules, stage 2.
MODPOST 11 modules
ERROR: "dax_pmem_compat_test" [tools/testing/nvdimm/test/nfit_test.ko] undefined!

Fix the problem by calling dax_pmem_compat_test() only if the kernel has the required functionality.

---
 tools/testing/nvdimm/test/nfit.c | 2 ++
 1 file changed, 2 insertions(+)
tools/testing/nvdimm/test/nfit.c +2 -0
index bf6422a6af7ff..a8ee5c4d41ebb 100644
@@ -3164,7 +3164,9 @@ static __init int nfit_test_init(void)
 	mcsafe_test();
 	dax_pmem_test();
 	dax_pmem_core_test();
+#ifdef CONFIG_DEV_DAX_PMEM_COMPAT
 	dax_pmem_compat_test();
+#endif
 
 	nfit_test_setup(nfit_test_lookup, nfit_test_evaluate_dsm);
 
Signature · 1 line
2.20.1
author maintainer
#2 lore raw

PATCH 3/78 watchdog: reset last_hw_keepalive time at start

In reply to Sasha Levin

From: Tero Kristo <[email protected]>

[ Upstream commit 982bb70517aef2225bad1d802887b733db492cc0 ]

Currently the watchdog core does not initialize the last_hw_keepalive time during watchdog startup. This will cause the watchdog to be pinged immediately if enough time has passed from the system boot-up time, and some types of watchdogs like K3 RTI does not like this.

To avoid the issue, setup the last_hw_keepalive time during watchdog startup.

---
 drivers/watchdog/watchdog_dev.c | 1 +
 1 file changed, 1 insertion(+)
drivers/watchdog/watchdog_dev.c +1 -0
index ce04edc69e5f0..c4147e93aa7d4 100644
@@ -282,6 +282,7 @@ static int watchdog_start(struct watchdog_device *wdd)
 	if (err == 0) {
 		set_bit(WDOG_ACTIVE, &wdd->status);
 		wd_data->last_keepalive = started_at;
+		wd_data->last_hw_keepalive = started_at;
 		watchdog_update_worker(wdd);
 	}
 
Signature · 1 line
2.20.1
author maintainer
#3 lore raw

PATCH 4/78 ovl: fix value of i_ino for lower hardlink corner case

In reply to Sasha Levin

From: Amir Goldstein <[email protected]>

[ Upstream commit 300b124fcf6ad2cd99a7b721e0f096785e0a3134 ]

Commit 6dde1e42f497 ("ovl: make i_ino consistent with st_ino in more cases"), relaxed the condition nfs_export=on in order to set the value of i_ino to xino map of real ino.

Specifically, it also relaxed the pre-condition that index=on for consistent i_ino. This opened the corner case of lower hardlink in ovl_get_inode(), which calls ovl_fill_inode() with ino=0 and then ovl_init_inode() is called to set i_ino to lower real ino without the xino mapping.

Pass the correct values of ino;fsid in this case to ovl_fill_inode(), so it can initialize i_ino correctly.

Fixes 6dde1e42f497 ("ovl: make i_ino consistent with st_ino in more ...")
---
 fs/overlayfs/inode.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
fs/overlayfs/inode.c +3 -1
index b045cf1826fc4..bb980721502dd 100644
@@ -881,7 +881,7 @@ struct inode *ovl_get_inode(struct super_block *sb,
 	struct dentry *lowerdentry = lowerpath ? lowerpath->dentry : NULL;
 	bool bylower = ovl_hash_bylower(sb, upperdentry, lowerdentry,
 					oip->index);
-	int fsid = bylower ? oip->lowerpath->layer->fsid : 0;
+	int fsid = bylower ? lowerpath->layer->fsid : 0;
 	bool is_dir, metacopy = false;
 	unsigned long ino = 0;
 	int err = oip->newinode ? -EEXIST : -ENOMEM;
@@ -931,6 +931,8 @@ struct inode *ovl_get_inode(struct super_block *sb,
 			err = -ENOMEM;
 			goto out_err;
 		}
+		ino = realinode->i_ino;
+		fsid = lowerpath->layer->fsid;
 	}
 	ovl_fill_inode(inode, realinode->i_mode, realinode->i_rdev, ino, fsid);
 	ovl_inode_init(inode, upperdentry, lowerdentry, oip->lowerdata);
Signature · 1 line
2.20.1
author maintainer
#4 lore raw

PATCH 5/78 iommu/vt-d: Fix page request descriptor size

In reply to Sasha Levin

From: Jacob Pan <[email protected]>

[ Upstream commit 52355fb1919ef7ed9a38e0f3de6e928de1f57217 ]

Intel VT-d might support PRS (Page Reqest Support) when it's running in the scalable mode. Each page request descriptor occupies 32 bytes and is 32-bytes aligned. The page request descriptor offset mask should be 32-bytes aligned.

Fixes 5b438f4ba315d ("iommu/vt-d: Support page request in scalable mode")
---
 drivers/iommu/intel-svm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/iommu/intel-svm.c +1 -1
index 518d0b2d12afd..5313d043a6cf2 100644
@@ -502,7 +502,7 @@ struct page_req_dsc {
 	u64 priv_data[2];
 };
 
-#define PRQ_RING_MASK ((0x1000 << PRQ_ORDER) - 0x10)
+#define PRQ_RING_MASK	((0x1000 << PRQ_ORDER) - 0x20)
 
 static bool access_error(struct vm_area_struct *vma, struct page_req_dsc *req)
 {
Signature · 6 lines
2.20.1

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
author maintainer
#5 lore raw

PATCH 6/78 acpi/nfit: improve bounds checking for 'func'

In reply to Sasha Levin

From: Dan Carpenter <[email protected]>

[ Upstream commit 01091c496f920e634ea84b689f480c39016752a8 ]

The 'func' variable can come from the user in the __nd_ioctl(). If it's too high then the (1 << func) shift in acpi_nfit_clear_to_send() is undefined. In acpi_nfit_ctl() we pass 'func' to test_bit(func, &dsm_mask) which could result in an out of bounds access.

To fix these issues, I introduced the NVDIMM_CMD_MAX (31) define and updated nfit_dsm_revid() to use that define as well instead of magic numbers.

Fixes 11189c1089da ("acpi/nfit: Fix command-supported detection")
---
 drivers/acpi/nfit/core.c | 10 ++++++----
 drivers/acpi/nfit/nfit.h |  1 +
 2 files changed, 7 insertions(+), 4 deletions(-)
drivers/acpi/nfit/core.c +6 -4
index 14e68f202f810..12d980aafc5ff 100644
@@ -360,7 +360,7 @@ static union acpi_object *acpi_label_info(acpi_handle handle)
 
 static u8 nfit_dsm_revid(unsigned family, unsigned func)
 {
-	static const u8 revid_table[NVDIMM_FAMILY_MAX+1][32] = {
+	static const u8 revid_table[NVDIMM_FAMILY_MAX+1][NVDIMM_CMD_MAX+1] = {
 		[NVDIMM_FAMILY_INTEL] = {
 			[NVDIMM_INTEL_GET_MODES] = 2,
 			[NVDIMM_INTEL_GET_FWINFO] = 2,
@@ -386,7 +386,7 @@ static u8 nfit_dsm_revid(unsigned family, unsigned func)
 
 	if (family > NVDIMM_FAMILY_MAX)
 		return 0;
-	if (func > 31)
+	if (func > NVDIMM_CMD_MAX)
 		return 0;
 	id = revid_table[family][func];
 	if (id == 0)
@@ -492,7 +492,8 @@ int acpi_nfit_ctl(struct nvdimm_bus_descriptor *nd_desc, struct nvdimm *nvdimm,
 	 * Check for a valid command.  For ND_CMD_CALL, we also have to
 	 * make sure that the DSM function is supported.
 	 */
-	if (cmd == ND_CMD_CALL && !test_bit(func, &dsm_mask))
+	if (cmd == ND_CMD_CALL &&
+	    (func > NVDIMM_CMD_MAX || !test_bit(func, &dsm_mask)))
 		return -ENOTTY;
 	else if (!test_bit(cmd, &cmd_mask))
 		return -ENOTTY;
@@ -3499,7 +3500,8 @@ static int acpi_nfit_clear_to_send(struct nvdimm_bus_descriptor *nd_desc,
 	if (nvdimm && cmd == ND_CMD_CALL &&
 			call_pkg->nd_family == NVDIMM_FAMILY_INTEL) {
 		func = call_pkg->nd_command;
-		if ((1 << func) & NVDIMM_INTEL_SECURITY_CMDMASK)
+		if (func > NVDIMM_CMD_MAX ||
+		    (1 << func) & NVDIMM_INTEL_SECURITY_CMDMASK)
 			return -EOPNOTSUPP;
 	}
 
drivers/acpi/nfit/nfit.h +1 -0
index 24241941181ce..b317f4043705f 100644
@@ -34,6 +34,7 @@
 		| ACPI_NFIT_MEM_NOT_ARMED | ACPI_NFIT_MEM_MAP_FAILED)
 
 #define NVDIMM_FAMILY_MAX NVDIMM_FAMILY_HYPERV
+#define NVDIMM_CMD_MAX 31
 
 #define NVDIMM_STANDARD_CMDMASK \
 (1 << ND_CMD_SMART | 1 << ND_CMD_SMART_THRESHOLD | 1 << ND_CMD_DIMM_FLAGS \
Signature · 4 lines
2.20.1
_______________________________________________
Linux-nvdimm mailing list -- [email protected]
To unsubscribe send an email to [email protected]
author maintainer
#6 lore raw

PATCH 7/78 iommu/vt-d: Fix mm reference leak

In reply to Sasha Levin

From: Jacob Pan <[email protected]>

[ Upstream commit 902baf61adf6b187f0a6b789e70d788ea71ff5bc ]

Move canonical address check before mmget_not_zero() to avoid mm reference leak.

Fixes 9d8c3af31607 ("iommu/vt-d: IOMMU Page Request needs to check if address is canonical.")
---
 drivers/iommu/intel-svm.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/iommu/intel-svm.c +4 -3
index 5313d043a6cf2..1d3816cd65d57 100644
@@ -583,14 +583,15 @@ static irqreturn_t prq_event_thread(int irq, void *d)
 		 * any faults on kernel addresses. */
 		if (!svm->mm)
 			goto bad_req;
-		/* If the mm is already defunct, don't handle faults. */
-		if (!mmget_not_zero(svm->mm))
-			goto bad_req;
 
 		/* If address is not canonical, return invalid response */
 		if (!is_canonical_address(address))
 			goto bad_req;
 
+		/* If the mm is already defunct, don't handle faults. */
+		if (!mmget_not_zero(svm->mm))
+			goto bad_req;
+
 		down_read(&svm->mm->mmap_sem);
 		vma = find_extend_vma(svm->mm, address);
 		if (!vma || address < vma->vm_start)
Signature · 1 line
2.20.1
author maintainer
#7 lore raw

PATCH 8/78 scsi: lpfc: Fix kasan slab-out-of-bounds error in lpfc_unreg_login

In reply to Sasha Levin

From: James Smart <[email protected]>

[ Upstream commit 38503943c89f0bafd9e3742f63f872301d44cbea ]

The following kasan bug was called out:

 BUG: KASAN: slab-out-of-bounds in lpfc_unreg_login+0x7c/0xc0 [lpfc]
 Read of size 2 at addr ffff889fc7c50a22 by task lpfc_worker_3/6676
 ...
 Call Trace:
 dump_stack+0x96/0xe0
 ? lpfc_unreg_login+0x7c/0xc0 [lpfc]
 print_address_description.constprop.6+0x1b/0x220
 ? lpfc_unreg_login+0x7c/0xc0 [lpfc]
 ? lpfc_unreg_login+0x7c/0xc0 [lpfc]
 __kasan_report.cold.9+0x37/0x7c
 ? lpfc_unreg_login+0x7c/0xc0 [lpfc]
 kasan_report+0xe/0x20
 lpfc_unreg_login+0x7c/0xc0 [lpfc]
 lpfc_sli_def_mbox_cmpl+0x334/0x430 [lpfc]
 ...

When processing the completion of a "Reg Rpi" login mailbox command in lpfc_sli_def_mbox_cmpl, a call may be made to lpfc_unreg_login. The vpi is extracted from the completing mailbox context and passed as an input for the next. However, the vpi stored in the mailbox command context is an absolute vpi, which for SLI4 represents both base + offset. When used with a non-zero base component, (function id > 0) this results in an out-of-range access beyond the allocated phba->vpi_ids array.

Fix by subtracting the function's base value to get an accurate vpi number.

---
 drivers/scsi/lpfc/lpfc_sli.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/scsi/lpfc/lpfc_sli.c +2 -0
index e2cec1f6e659b..8836c5682c8d5 100644
@@ -2480,6 +2480,8 @@ lpfc_sli_def_mbox_cmpl(struct lpfc_hba *phba, LPFC_MBOXQ_t *pmb)
 	    !pmb->u.mb.mbxStatus) {
 		rpi = pmb->u.mb.un.varWords[0];
 		vpi = pmb->u.mb.un.varRegLogin.vpi;
+		if (phba->sli_rev == LPFC_SLI_REV4)
+			vpi -= phba->sli4_hba.max_cfg_param.vpi_base;
 		lpfc_unreg_login(phba, vpi, rpi, pmb);
 		pmb->vport = vport;
 		pmb->mbox_cmpl = lpfc_sli_def_mbox_cmpl;
Signature · 1 line
2.20.1
author maintainer
#8 lore raw

PATCH 9/78 scsi: lpfc: Fix crash after handling a pci error

In reply to Sasha Levin

From: James Smart <[email protected]>

[ Upstream commit 4cd70891308dfb875ef31060c4a4aa8872630a2e ]

Injecting EEH on a 32GB card is causing kernel oops

The pci error handler is doing an IO flush and the offline code is also doing an IO flush. When the 1st flush is complete the hdwq is destroyed (freed), yet the second flush accesses the hdwq and crashes.

Added a check in lpfc_sli4_fush_io_rings to check both the HBA_IOQ_FLUSH flag and the hdwq pointer to see if it is already set and not already freed.

---
 drivers/scsi/lpfc/lpfc_sli.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/scsi/lpfc/lpfc_sli.c +5 -0
index 8836c5682c8d5..4e04270cc5c4f 100644
@@ -4012,6 +4012,11 @@ lpfc_sli_flush_io_rings(struct lpfc_hba *phba)
 	struct lpfc_iocbq *piocb, *next_iocb;
 
 	spin_lock_irq(&phba->hbalock);
+	if (phba->hba_flag & HBA_IOQ_FLUSH ||
+	    !phba->sli4_hba.hdwq) {
+		spin_unlock_irq(&phba->hbalock);
+		return;
+	}
 	/* Indicate the I/O queues are flushed */
 	phba->hba_flag |= HBA_IOQ_FLUSH;
 	spin_unlock_irq(&phba->hbalock);
Signature · 1 line
2.20.1
author maintainer
#9 lore raw

PATCH 10/78 scsi: lpfc: Fix crash in target side cable pulls hitting WAIT_FOR_UNREG

In reply to Sasha Levin

From: James Smart <[email protected]>

[ Upstream commit 807e7353d8a7105ce884d22b0dbc034993c6679c ]

Kernel is crashing with the following stacktrace:

  BUG: unable to handle kernel NULL pointer dereference at
    00000000000005bc
  IP: lpfc_nvme_register_port+0x1a8/0x3a0 [lpfc]
  ...
  Call Trace:
  lpfc_nlp_state_cleanup+0x2b2/0x500 [lpfc]
  lpfc_nlp_set_state+0xd7/0x1a0 [lpfc]
  lpfc_cmpl_prli_prli_issue+0x1f7/0x450 [lpfc]
  lpfc_disc_state_machine+0x7a/0x1e0 [lpfc]
  lpfc_cmpl_els_prli+0x16f/0x1e0 [lpfc]
  lpfc_sli_sp_handle_rspiocb+0x5b2/0x690 [lpfc]
  lpfc_sli_handle_slow_ring_event_s4+0x182/0x230 [lpfc]
  lpfc_do_work+0x87f/0x1570 [lpfc]
  kthread+0x10d/0x130
  ret_from_fork+0x35/0x40

During target side fault injections, it is possible to hit the NLP_WAIT_FOR_UNREG case in lpfc_nvme_remoteport_delete. A prior commit fixed a rebind and delete race condition, but called lpfc_nlp_put unconditionally. This triggered a deletion and the crash.

Fix by movng nlp_put to inside the NLP_WAIT_FOR_UNREG case, where the nlp will be being unregistered/removed. Leave the reference if the flag isn't set.

Fixes b15bd3e6212e ("scsi: lpfc: Fix nvme remoteport registration race conditions")
---
 drivers/scsi/lpfc/lpfc_nvme.c | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)
drivers/scsi/lpfc/lpfc_nvme.c +8 -6
index 8e0f03ef346b6..c475ae0f2f515 100644
@@ -342,13 +342,15 @@ lpfc_nvme_remoteport_delete(struct nvme_fc_remote_port *remoteport)
 	if (ndlp->upcall_flags & NLP_WAIT_FOR_UNREG) {
 		ndlp->nrport = NULL;
 		ndlp->upcall_flags &= ~NLP_WAIT_FOR_UNREG;
-	}
-	spin_unlock_irq(&vport->phba->hbalock);
+		spin_unlock_irq(&vport->phba->hbalock);
 
-	/* Remove original register reference. The host transport
-	 * won't reference this rport/remoteport any further.
-	 */
-	lpfc_nlp_put(ndlp);
+		/* Remove original register reference. The host transport
+		 * won't reference this rport/remoteport any further.
+		 */
+		lpfc_nlp_put(ndlp);
+	} else {
+		spin_unlock_irq(&vport->phba->hbalock);
+	}
 
  rport_err:
 	return;
Signature · 1 line
2.20.1
author maintainer
#10 lore raw

PATCH 11/78 scsi: libfc: If PRLI rejected, move rport to PLOGI state

In reply to Sasha Levin

From: Javed Hasan <[email protected]>

[ Upstream commit 45e544bfdab2014d11c7595b8ccc3c4715a09015 ]

If PRLI reject code indicates "rejected status", move rport state machine back to PLOGI state.

---
 drivers/scsi/libfc/fc_rport.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
drivers/scsi/libfc/fc_rport.c +7 -1
index da6e97d8dc3bb..6bb8917b99a19 100644
@@ -1208,9 +1208,15 @@ static void fc_rport_prli_resp(struct fc_seq *sp, struct fc_frame *fp,
 		rjt = fc_frame_payload_get(fp, sizeof(*rjt));
 		if (!rjt)
 			FC_RPORT_DBG(rdata, "PRLI bad response\n");
-		else
+		else {
 			FC_RPORT_DBG(rdata, "PRLI ELS rejected, reason %x expl %x\n",
 				     rjt->er_reason, rjt->er_explan);
+			if (rjt->er_reason == ELS_RJT_UNAB &&
+			    rjt->er_explan == ELS_EXPL_PLOGI_REQD) {
+				fc_rport_enter_plogi(rdata);
+				goto out;
+			}
+		}
 		fc_rport_error_retry(rdata, FC_EX_ELS_RJT);
 	}
 
Signature · 1 line
2.20.1
author maintainer
#11 lore raw

PATCH 12/78 ceph: return ceph_mdsc_do_request() errors from __get_parent()

In reply to Sasha Levin

From: Qiujun Huang <[email protected]>

[ Upstream commit c6d50296032f0b97473eb2e274dc7cc5d0173847 ]

Return the error returned by ceph_mdsc_do_request(). Otherwise, r_target_inode ends up being NULL this ends up returning ENOENT regardless of the error.

---
 fs/ceph/export.c | 5 +++++
 1 file changed, 5 insertions(+)
fs/ceph/export.c +5 -0
index b6bfa94332c30..79dc06881e78e 100644
@@ -315,6 +315,11 @@ static struct dentry *__get_parent(struct super_block *sb,
 
 	req->r_num_caps = 1;
 	err = ceph_mdsc_do_request(mdsc, NULL, req);
+	if (err) {
+		ceph_mdsc_put_request(req);
+		return ERR_PTR(err);
+	}
+
 	inode = req->r_target_inode;
 	if (inode)
 		ihold(inode);
Signature · 1 line
2.20.1
author maintainer
#12 lore raw

PATCH 13/78 ceph: don't skip updating wanted caps when cap is stale

In reply to Sasha Levin

From: "Yan, Zheng" <[email protected]>

[ Upstream commit 0aa971b6fd3f92afef6afe24ef78d9bb14471519 ]

1. try_get_cap_refs() fails to get caps and finds that mds_wanted does not include what it wants. It returns -ESTALE.
2. ceph_get_caps() calls ceph_renew_caps(). ceph_renew_caps() finds that inode has cap, so it calls ceph_check_caps().
3. ceph_check_caps() finds that issued caps (without checking if it's stale) already includes caps wanted by open file, so it skips updating wanted caps.

Above events can cause an infinite loop inside ceph_get_caps().

---
 fs/ceph/caps.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
fs/ceph/caps.c +6 -2
index f5a38910a82bf..703945cce0e5d 100644
@@ -1976,8 +1976,12 @@ void ceph_check_caps(struct ceph_inode_info *ci, int flags,
 		}
 
 		/* want more caps from mds? */
-		if (want & ~(cap->mds_wanted | cap->issued))
-			goto ack;
+		if (want & ~cap->mds_wanted) {
+			if (want & ~(cap->mds_wanted | cap->issued))
+				goto ack;
+			if (!__cap_is_valid(cap))
+				goto ack;
+		}
 
 		/* things we might delay */
 		if ((cap->issued & ~retain) == 0)
Signature · 1 line
2.20.1
author maintainer
#13 lore raw

PATCH 14/78 pwm: rcar: Fix late Runtime PM enablement

In reply to Sasha Levin

From: Geert Uytterhoeven <[email protected]>

[ Upstream commit 1451a3eed24b5fd6a604683f0b6995e0e7e16c79 ]

Runtime PM should be enabled before calling pwmchip_add(), as PWM users can appear immediately after the PWM chip has been added. Likewise, Runtime PM should be disabled after the removal of the PWM chip.

Fixes ed6c1476bf7f16d5 ("pwm: Add support for R-Car PWM Timer")
---
 drivers/pwm/pwm-rcar.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
drivers/pwm/pwm-rcar.c +7 -3
index 852eb2347954d..b98ec8847b488 100644
@@ -228,24 +228,28 @@ static int rcar_pwm_probe(struct platform_device *pdev)
 	rcar_pwm->chip.base = -1;
 	rcar_pwm->chip.npwm = 1;
 
+	pm_runtime_enable(&pdev->dev);
+
 	ret = pwmchip_add(&rcar_pwm->chip);
 	if (ret < 0) {
 		dev_err(&pdev->dev, "failed to register PWM chip: %d\n", ret);
+		pm_runtime_disable(&pdev->dev);
 		return ret;
 	}
 
-	pm_runtime_enable(&pdev->dev);
-
 	return 0;
 }
 
 static int rcar_pwm_remove(struct platform_device *pdev)
 {
 	struct rcar_pwm_chip *rcar_pwm = platform_get_drvdata(pdev);
+	int ret;
+
+	ret = pwmchip_remove(&rcar_pwm->chip);
 
 	pm_runtime_disable(&pdev->dev);
 
-	return pwmchip_remove(&rcar_pwm->chip);
+	return ret;
 }
 
 static const struct of_device_id rcar_pwm_of_table[] = {
Signature · 1 line
2.20.1
author maintainer
#14 lore raw

PATCH 15/78 nvme-tcp: fix possible crash in write_zeroes processing

In reply to Sasha Levin

From: Sagi Grimberg <[email protected]>

[ Upstream commit 25e5cb780e62bde432b401f312bb847edc78b432 ]

We cannot look at blk_rq_payload_bytes without first checking that the request has a mappable physical segments first (e.g. blk_rq_nr_phys_segments(rq) != 0) and only then to take the request payload bytes. This caused us to send a wrong sgl to the target or even dereference a non-existing buffer in case we actually got to the data send sequence (if it was in-capsule).

---
 drivers/nvme/host/tcp.c | 13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)
drivers/nvme/host/tcp.c +7 -6
index 244984420b41b..11e84ed4de361 100644
@@ -164,16 +164,14 @@ static inline bool nvme_tcp_async_req(struct nvme_tcp_request *req)
 static inline bool nvme_tcp_has_inline_data(struct nvme_tcp_request *req)
 {
 	struct request *rq;
-	unsigned int bytes;
 
 	if (unlikely(nvme_tcp_async_req(req)))
 		return false; /* async events don't have a request */
 
 	rq = blk_mq_rq_from_pdu(req);
-	bytes = blk_rq_payload_bytes(rq);
 
-	return rq_data_dir(rq) == WRITE && bytes &&
-		bytes <= nvme_tcp_inline_data_size(req->queue);
+	return rq_data_dir(rq) == WRITE && req->data_len &&
+		req->data_len <= nvme_tcp_inline_data_size(req->queue);
 }
 
 static inline struct page *nvme_tcp_req_cur_page(struct nvme_tcp_request *req)
@@ -2090,7 +2088,9 @@ static blk_status_t nvme_tcp_map_data(struct nvme_tcp_queue *queue,
 
 	c->common.flags |= NVME_CMD_SGL_METABUF;
 
-	if (rq_data_dir(rq) == WRITE && req->data_len &&
+	if (!blk_rq_nr_phys_segments(rq))
+		nvme_tcp_set_sg_null(c);
+	else if (rq_data_dir(rq) == WRITE &&
 	    req->data_len <= nvme_tcp_inline_data_size(queue))
 		nvme_tcp_set_sg_inline(queue, c, req->data_len);
 	else
@@ -2117,7 +2117,8 @@ static blk_status_t nvme_tcp_setup_cmd_pdu(struct nvme_ns *ns,
 	req->data_sent = 0;
 	req->pdu_len = 0;
 	req->pdu_sent = 0;
-	req->data_len = blk_rq_payload_bytes(rq);
+	req->data_len = blk_rq_nr_phys_segments(rq) ?
+				blk_rq_payload_bytes(rq) : 0;
 	req->curr_bio = rq->bio;
 
 	if (rq_data_dir(rq) == WRITE &&
Signature · 7 lines
2.20.1


_______________________________________________
linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme
author maintainer
#15 lore raw

PATCH 16/78 ASoC: dpcm: allow start or stop during pause for backend

In reply to Sasha Levin

From: 이경택 <[email protected]>

[ Upstream commit 21fca8bdbb64df1297e8c65a746c4c9f4a689751 ]

soc_compr_trigger_fe() allows start or stop after pause_push. In dpcm_be_dai_trigger(), however, only pause_release is allowed command after pause_push.
So, start or stop after pause in compress offload is always returned as error if the compress offload is used with dpcm. To fix the problem, SND_SOC_DPCM_STATE_PAUSED should be allowed for start or stop command.

---
 sound/soc/soc-pcm.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
sound/soc/soc-pcm.c +4 -2
index d978df95c5c65..cc4e9aa80fb0d 100644
@@ -2222,7 +2222,8 @@ int dpcm_be_dai_trigger(struct snd_soc_pcm_runtime *fe, int stream,
 		switch (cmd) {
 		case SNDRV_PCM_TRIGGER_START:
 			if ((be->dpcm[stream].state != SND_SOC_DPCM_STATE_PREPARE) &&
-			    (be->dpcm[stream].state != SND_SOC_DPCM_STATE_STOP))
+			    (be->dpcm[stream].state != SND_SOC_DPCM_STATE_STOP) &&
+			    (be->dpcm[stream].state != SND_SOC_DPCM_STATE_PAUSED))
 				continue;
 
 			ret = dpcm_do_trigger(dpcm, be_substream, cmd);
@@ -2252,7 +2253,8 @@ int dpcm_be_dai_trigger(struct snd_soc_pcm_runtime *fe, int stream,
 			be->dpcm[stream].state = SND_SOC_DPCM_STATE_START;
 			break;
 		case SNDRV_PCM_TRIGGER_STOP:
-			if (be->dpcm[stream].state != SND_SOC_DPCM_STATE_START)
+			if ((be->dpcm[stream].state != SND_SOC_DPCM_STATE_START) &&
+			    (be->dpcm[stream].state != SND_SOC_DPCM_STATE_PAUSED))
 				continue;
 
 			if (!snd_soc_dpcm_can_be_free_stop(fe, be, stream))
Signature · 1 line
2.20.1
author maintainer
#16 lore raw

PATCH 17/78 scsi: iscsi: Report unbind session event when the target has been removed

In reply to Sasha Levin

From: Wu Bo <[email protected]>

[ Upstream commit 13e60d3ba287d96eeaf1deaadba51f71578119a3 ]

If the daemon is restarted or crashes while logging out of a session, the unbind session event sent by the kernel is not processed and is lost. When the daemon starts again, the session can't be unbound because the daemon is waiting for the event message. However, the kernel has already logged out and the event will not be resent.

When iscsid restart is complete, logout session reports error:

Logging out of session [sid: 6, target: iqn.xxxxx, portal: xx.xx.xx.xx,3260]
iscsiadm: Could not logout of [sid: 6, target: iscsiadm -m node iqn.xxxxx, portal: xx.xx.xx.xx,3260].
iscsiadm: initiator reported error (9 - internal error)
iscsiadm: Could not logout of all requested sessions

Make sure the unbind event is emitted.

[mkp: commit desc and applied by hand since patch was mangled]

---
 drivers/scsi/scsi_transport_iscsi.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/scsi/scsi_transport_iscsi.c +3 -1
index 271afea654e2b..a5c78b38d3022 100644
@@ -2012,7 +2012,7 @@ static void __iscsi_unbind_session(struct work_struct *work)
 	if (session->target_id == ISCSI_MAX_TARGET) {
 		spin_unlock_irqrestore(&session->lock, flags);
 		mutex_unlock(&ihost->mutex);
-		return;
+		goto unbind_session_exit;
 	}
 
 	target_id = session->target_id;
@@ -2024,6 +2024,8 @@ static void __iscsi_unbind_session(struct work_struct *work)
 		ida_simple_remove(&iscsi_sess_ida, target_id);
 
 	scsi_remove_target(&session->dev);
+
+unbind_session_exit:
 	iscsi_session_event(session, ISCSI_KEVENT_UNBIND_SESSION);
 	ISCSI_DBG_TRANS_SESSION(session, "Completed target removal\n");
 }
Signature · 1 line
2.20.1
author maintainer
#17 lore raw

PATCH 18/78 tools/test/nvdimm: Fix out of tree build

In reply to Sasha Levin

From: Santosh Sivaraj <[email protected]>

[ Upstream commit 1f776799628139d0da47e710ad86eb58d987ff66 ]

Out of tree build using

make M=tools/test/nvdimm O=/tmp/build -C /tmp/build

fails with the following error

make: Entering directory '/tmp/build'
CC [M] tools/testing/nvdimm/test/nfit.o
linux/tools/testing/nvdimm/test/nfit.c:19:10: fatal error: nd-core.h: No such file or directory
19 | #include <nd-core.h>

      |          ^~~~~~~~~~~

compilation terminated.

That is because the kbuild file uses $(src) which points to tools/testing/nvdimm, $(srctree) correctly points to root of the linux source tree.

---
 tools/testing/nvdimm/Kbuild      | 4 ++--
 tools/testing/nvdimm/test/Kbuild | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)
tools/testing/nvdimm/Kbuild +2 -2
index c4a9196d794c9..cb80d3b811792 100644
@@ -21,8 +21,8 @@ DRIVERS := ../../../drivers
 NVDIMM_SRC := $(DRIVERS)/nvdimm
 ACPI_SRC := $(DRIVERS)/acpi/nfit
 DAX_SRC := $(DRIVERS)/dax
-ccflags-y := -I$(src)/$(NVDIMM_SRC)/
-ccflags-y += -I$(src)/$(ACPI_SRC)/
+ccflags-y := -I$(srctree)/drivers/nvdimm/
+ccflags-y += -I$(srctree)/drivers/acpi/nfit/
 
 obj-$(CONFIG_LIBNVDIMM) += libnvdimm.o
 obj-$(CONFIG_BLK_DEV_PMEM) += nd_pmem.o
tools/testing/nvdimm/test/Kbuild +2 -2
index fb3c3d7cdb9bd..75baebf8f4ba1 100644
@@ -1,6 +1,6 @@
 # SPDX-License-Identifier: GPL-2.0
-ccflags-y := -I$(src)/../../../../drivers/nvdimm/
-ccflags-y += -I$(src)/../../../../drivers/acpi/nfit/
+ccflags-y := -I$(srctree)/drivers/nvdimm/
+ccflags-y += -I$(srctree)/drivers/acpi/nfit/
 
 obj-m += nfit_test.o
 obj-m += nfit_test_iomap.o
Signature · 4 lines
2.20.1
_______________________________________________
Linux-nvdimm mailing list -- [email protected]
To unsubscribe send an email to [email protected]
author maintainer
#18 lore raw

PATCH 19/78 ASoC: Intel: atom: Take the drv->lock mutex before calling sst_send_slot_map()

In reply to Sasha Levin

From: Hans de Goede <[email protected]>

[ Upstream commit 81630dc042af998b9f58cd8e2c29dab9777ea176 ]

sst_send_slot_map() uses sst_fill_and_send_cmd_unlocked() because in some places it is called with the drv->lock mutex already held.

So it must always be called with the mutex locked. This commit adds missing locking in the sst_set_be_modules() code-path.

---
 sound/soc/intel/atom/sst-atom-controls.c | 2 ++
 1 file changed, 2 insertions(+)
sound/soc/intel/atom/sst-atom-controls.c +2 -0
index baef461a99f19..2c3798034b1de 100644
@@ -966,7 +966,9 @@ static int sst_set_be_modules(struct snd_soc_dapm_widget *w,
 	dev_dbg(c->dev, "Enter: widget=%s\n", w->name);
 
 	if (SND_SOC_DAPM_EVENT_ON(event)) {
+		mutex_lock(&drv->lock);
 		ret = sst_send_slot_map(drv);
+		mutex_unlock(&drv->lock);
 		if (ret)
 			return ret;
 		ret = sst_send_pipe_module_params(w, k);
Signature · 1 line
2.20.1
author maintainer
#19 lore raw

PATCH 20/78 nvme: fix deadlock caused by ANA update wrong locking

In reply to Sasha Levin

From: Sagi Grimberg <[email protected]>

[ Upstream commit 657f1975e9d9c880fa13030e88ba6cc84964f1db ]

The deadlock combines 4 flows in parallel:
- ns scanning (triggered from reconnect)
- request timeout
- ANA update (triggered from reconnect)
- I/O coming into the mpath device

(1) ns scanning triggers disk revalidation -> update disk info ->

    freeze queue -> but blocked, due to (2)

(2) timeout handler reference the g_usage_counter - > but blocks in the transport .timeout() handler, due to (3)

(3) the transport timeout handler (indirectly) calls nvme_stop_queue() -> which takes the (down_read) namespaces_rwsem - > but blocks, due to (4)

(4) ANA update takes the (down_write) namespaces_rwsem -> calls

    nvme_mpath_set_live() -> which synchronize the ns_head srcu
    (see commit 504db087aacc) -> but blocks, due to (5)

(5) I/O came into nvme_mpath_make_request -> took srcu_read_lock ->

    direct_make_request > blk_queue_enter -> but blocked, due to (1)

==> the request queue is under freeze -> deadlock.

The fix is making ANA update take a read lock as the namespaces list is not manipulated, it is just the ns and ns->head that are being updated (which is protected with the ns->head lock).

---
 drivers/nvme/host/multipath.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/nvme/host/multipath.c +2 -2
index aed6354cb2717..56caddeabb5e5 100644
@@ -510,7 +510,7 @@ static int nvme_update_ana_state(struct nvme_ctrl *ctrl,
 	if (!nr_nsids)
 		return 0;
 
-	down_write(&ctrl->namespaces_rwsem);
+	down_read(&ctrl->namespaces_rwsem);
 	list_for_each_entry(ns, &ctrl->namespaces, list) {
 		unsigned nsid = le32_to_cpu(desc->nsids[n]);
 
@@ -521,7 +521,7 @@ static int nvme_update_ana_state(struct nvme_ctrl *ctrl,
 		if (++n == nr_nsids)
 			break;
 	}
-	up_write(&ctrl->namespaces_rwsem);
+	up_read(&ctrl->namespaces_rwsem);
 	return 0;
 }
 
Signature · 7 lines
2.20.1


_______________________________________________
linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme
author maintainer
#20 lore raw

PATCH 21/78 drm/amd/display: Update stream adjust in dc_stream_adjust_vmin_vmax

In reply to Sasha Levin

From: Isabel Zhang <[email protected]>

[ Upstream commit 346d8a0a3c91888a412c2735d69daa09c00f0203 ]

[Why]
After v_total_min and max are updated in vrr structure, the changes are not reflected in stream adjust. When these values are read from stream adjust it does not reflect the actual state of the system.

[How]
Set stream adjust values equal to vrr adjust values after vrr adjust values are updated.

---
 drivers/gpu/drm/amd/display/dc/core/dc.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/gpu/drm/amd/display/dc/core/dc.c +2 -0
index 4704aac336c29..89bd0ba3db1df 100644
@@ -283,6 +283,8 @@ bool dc_stream_adjust_vmin_vmax(struct dc *dc,
 	int i = 0;
 	bool ret = false;
 
+	stream->adjust = *adjust;
+
 	for (i = 0; i < MAX_PIPES; i++) {
 		struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i];
 
Signature · 6 lines
2.20.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
author maintainer
#21 lore raw

PATCH 22/78 dma-direct: fix data truncation in dma_direct_get_required_mask()

In reply to Sasha Levin

From: Kishon Vijay Abraham I <[email protected]>

[ Upstream commit cdcda0d1f8f4ab84efe7cd9921c98364398aefd7 ]

The upper 32-bit physical address gets truncated inadvertently when dma_direct_get_required_mask() invokes phys_to_dma_direct(). This results in dma_addressing_limited() return incorrect value when used in platforms with LPAE enabled.
Fix it here by explicitly type casting 'max_pfn' to phys_addr_t in order to prevent overflow of intermediate value while evaluating '(max_pfn - 1) << PAGE_SHIFT'.

---
 kernel/dma/direct.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
kernel/dma/direct.c +2 -1
index 867fd72cb2605..0a093a675b632 100644
@@ -45,7 +45,8 @@ static inline dma_addr_t phys_to_dma_direct(struct device *dev,
 
 u64 dma_direct_get_required_mask(struct device *dev)
 {
-	u64 max_dma = phys_to_dma_direct(dev, (max_pfn - 1) << PAGE_SHIFT);
+	phys_addr_t phys = (phys_addr_t)(max_pfn - 1) << PAGE_SHIFT;
+	u64 max_dma = phys_to_dma_direct(dev, phys);
 
 	return (1ULL << (fls64(max_dma) - 1)) * 2 - 1;
 }
Signature · 6 lines
2.20.1

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
author maintainer
#22 lore raw

PATCH 23/78 dma-debug: fix displaying of dma allocation type

In reply to Sasha Levin

From: Grygorii Strashko <[email protected]>

[ Upstream commit 9bb50ed7470944238ec8e30a94ef096caf9056ee ]

The commit 2e05ea5cdc1a ("dma-mapping: implement dma_map_single_attrs using dma_map_page_attrs") removed "dma_debug_page" enum, but missed to update type2name string table. This causes incorrect displaying of dma allocation type.
Fix it by removing "page" string from type2name string table and switch to use named initializers.

Before (dma_alloc_coherent()):
k3-ringacc 4b800000.ringacc: scather-gather idx 2208 P=d1140000 N=d114 D=d1140000 L=40 DMA_BIDIRECTIONAL dma map error check not applicable k3-ringacc 4b800000.ringacc: scather-gather idx 2216 P=d1150000 N=d115 D=d1150000 L=40 DMA_BIDIRECTIONAL dma map error check not applicable

After:
k3-ringacc 4b800000.ringacc: coherent idx 2208 P=d1140000 N=d114 D=d1140000 L=40 DMA_BIDIRECTIONAL dma map error check not applicable k3-ringacc 4b800000.ringacc: coherent idx 2216 P=d1150000 N=d115 D=d1150000 L=40 DMA_BIDIRECTIONAL dma map error check not applicable

Fixes 2e05ea5cdc1a ("dma-mapping: implement dma_map_single_attrs using dma_map_page_attrs")
---
 kernel/dma/debug.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)
kernel/dma/debug.c +6 -3
index 4ad74f5987ea9..cb6425e52bf7a 100644
@@ -137,9 +137,12 @@ static const char *const maperr2str[] = {
 	[MAP_ERR_CHECKED] = "dma map error checked",
 };
 
-static const char *type2name[5] = { "single", "page",
-				    "scather-gather", "coherent",
-				    "resource" };
+static const char *type2name[] = {
+	[dma_debug_single] = "single",
+	[dma_debug_sg] = "scather-gather",
+	[dma_debug_coherent] = "coherent",
+	[dma_debug_resource] = "resource",
+};
 
 static const char *dir2name[4] = { "DMA_BIDIRECTIONAL", "DMA_TO_DEVICE",
 				   "DMA_FROM_DEVICE", "DMA_NONE" };
Signature · 6 lines
2.20.1

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
author maintainer
#23 lore raw

PATCH 24/78 kernel/gcov/fs.c: gcov_seq_next() should increase position index

In reply to Sasha Levin

From: Vasily Averin <[email protected]>

[ Upstream commit f4d74ef6220c1eda0875da30457bef5c7111ab06 ]

If seq_file .next function does not change position index, read after some lseek can generate unexpected output.

https://bugzilla.kernel.org/show_bug.cgi?id=206283

---
 kernel/gcov/fs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/gcov/fs.c +1 -1
index e5eb5ea7ea598..cc4ee482d3fba 100644
@@ -108,9 +108,9 @@ static void *gcov_seq_next(struct seq_file *seq, void *data, loff_t *pos)
 {
 	struct gcov_iterator *iter = data;
 
+	(*pos)++;
 	if (gcov_iter_next(iter))
 		return NULL;
-	(*pos)++;
 
 	return iter;
 }
Signature · 1 line
2.20.1
author maintainer
#24 lore raw

PATCH 25/78 selftests: kmod: fix handling test numbers above 9

In reply to Sasha Levin

From: Eric Biggers <[email protected]>

[ Upstream commit 6d573a07528308eb77ec072c010819c359bebf6e ]

get_test_count() and get_test_enabled() were broken for test numbers above 9 due to awk interpreting a field specification like '$0010' as octal rather than decimal. Fix it by stripping the leading zeroes.

---
 tools/testing/selftests/kmod/kmod.sh | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)
tools/testing/selftests/kmod/kmod.sh +9 -4
index 8b944cf042f6c..315a43111e046 100755
@@ -505,18 +505,23 @@ function test_num()
 	fi
 }
 
-function get_test_count()
+function get_test_data()
 {
 	test_num $1
-	TEST_DATA=$(echo $ALL_TESTS | awk '{print $'$1'}')
+	local field_num=$(echo $1 | sed 's/^0*//')
+	echo $ALL_TESTS | awk '{print $'$field_num'}'
+}
+
+function get_test_count()
+{
+	TEST_DATA=$(get_test_data $1)
 	LAST_TWO=${TEST_DATA#*:*}
 	echo ${LAST_TWO%:*}
 }
 
 function get_test_enabled()
 {
-	test_num $1
-	TEST_DATA=$(echo $ALL_TESTS | awk '{print $'$1'}')
+	TEST_DATA=$(get_test_data $1)
 	echo ${TEST_DATA#*:*:}
 }
 
Signature · 1 line
2.20.1
author maintainer
#25 lore raw

PATCH 26/78 ipc/util.c: sysvipc_find_ipc() should increase position index

In reply to Sasha Levin

From: Vasily Averin <[email protected]>

[ Upstream commit 89163f93c6f969da5811af5377cc10173583123b ]

If seq_file .next function does not change position index, read after some lseek can generate unexpected output.

https://bugzilla.kernel.org/show_bug.cgi?id=206283

---
 ipc/util.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
ipc/util.c +1 -1
index d126d156efc64..594871610d454 100644
@@ -764,13 +764,13 @@ static struct kern_ipc_perm *sysvipc_find_ipc(struct ipc_ids *ids, loff_t pos,
 			total++;
 	}
 
+	*new_pos = pos + 1;
 	if (total >= ids->in_use)
 		return NULL;
 
 	for (; pos < ipc_mni; pos++) {
 		ipc = idr_find(&ids->ipcs_idr, pos);
 		if (ipc != NULL) {
-			*new_pos = pos + 1;
 			rcu_read_lock();
 			ipc_lock_object(ipc);
 			return ipc;
Signature · 1 line
2.20.1
author maintainer
#26 lore raw

PATCH 27/78 kconfig: qconf: Fix a few alignment issues

In reply to Sasha Levin

From: Mauro Carvalho Chehab <[email protected]>

[ Upstream commit 60969f02f07ae1445730c7b293c421d179da729c ]

There are a few items with wrong alignments. Solve them.

---
 scripts/kconfig/qconf.cc | 13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)
scripts/kconfig/qconf.cc +7 -6
index 82773cc35d356..0f8c77f847114 100644
@@ -627,7 +627,7 @@ void ConfigList::updateMenuList(ConfigItem *parent, struct menu* menu)
 			last = item;
 			continue;
 		}
-	hide:
+hide:
 		if (item && item->menu == child) {
 			last = parent->firstChild();
 			if (last == item)
@@ -692,7 +692,7 @@ void ConfigList::updateMenuList(ConfigList *parent, struct menu* menu)
 			last = item;
 			continue;
 		}
-	hide:
+hide:
 		if (item && item->menu == child) {
 			last = (ConfigItem*)parent->topLevelItem(0);
 			if (last == item)
@@ -1225,10 +1225,11 @@ QMenu* ConfigInfoView::createStandardContextMenu(const QPoint & pos)
 {
 	QMenu* popup = Parent::createStandardContextMenu(pos);
 	QAction* action = new QAction("Show Debug Info", popup);
-	  action->setCheckable(true);
-	  connect(action, SIGNAL(toggled(bool)), SLOT(setShowDebug(bool)));
-	  connect(this, SIGNAL(showDebugChanged(bool)), action, SLOT(setOn(bool)));
-	  action->setChecked(showDebug());
+
+	action->setCheckable(true);
+	connect(action, SIGNAL(toggled(bool)), SLOT(setShowDebug(bool)));
+	connect(this, SIGNAL(showDebugChanged(bool)), action, SLOT(setOn(bool)));
+	action->setChecked(showDebug());
 	popup->addSeparator();
 	popup->addAction(action);
 	return popup;
Signature · 1 line
2.20.1
author maintainer
#27 lore raw

PATCH 28/78 lib/raid6/test: fix build on distros whose /bin/sh is not bash

In reply to Sasha Levin

From: Masahiro Yamada <[email protected]>

[ Upstream commit 06bd48b6cd97ef3889b68c8e09014d81dbc463f1 ]

You can build a user-space test program for the raid6 library code, like this:

  $ cd lib/raid6/test
  $ make

The command in $(shell ...) function is evaluated by /bin/sh by default. (or, you can specify the shell by passing SHELL=<shell> from command line)

Currently '>&/dev/null' is used to sink both stdout and stderr. Because this code is bash-ism, it only works when /bin/sh is a symbolic link to bash (this is the case on RHEL etc.)

This does not work on Ubuntu where /bin/sh is a symbolic link to dash.

I see lots of

/bin/sh: 1: Syntax error: Bad fd number

and

warning "your version of binutils lacks ... support"

Replace it with portable '>/dev/null 2>&1'.

---
 lib/raid6/test/Makefile | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
lib/raid6/test/Makefile +3 -3
index 3ab8720aa2f84..b9e6c3648be1a 100644
@@ -35,13 +35,13 @@ endif
 ifeq ($(IS_X86),yes)
         OBJS   += mmx.o sse1.o sse2.o avx2.o recov_ssse3.o recov_avx2.o avx512.o recov_avx512.o
         CFLAGS += $(shell echo "pshufb %xmm0, %xmm0" |		\
-                    gcc -c -x assembler - >&/dev/null &&	\
+                    gcc -c -x assembler - >/dev/null 2>&1 &&	\
                     rm ./-.o && echo -DCONFIG_AS_SSSE3=1)
         CFLAGS += $(shell echo "vpbroadcastb %xmm0, %ymm1" |	\
-                    gcc -c -x assembler - >&/dev/null &&	\
+                    gcc -c -x assembler - >/dev/null 2>&1 &&	\
                     rm ./-.o && echo -DCONFIG_AS_AVX2=1)
 	CFLAGS += $(shell echo "vpmovm2b %k1, %zmm5" |          \
-		    gcc -c -x assembler - >&/dev/null &&        \
+		    gcc -c -x assembler - >/dev/null 2>&1 &&	\
 		    rm ./-.o && echo -DCONFIG_AS_AVX512=1)
 else ifeq ($(HAS_NEON),yes)
         OBJS   += neon.o neon1.o neon2.o neon4.o neon8.o recov_neon.o recov_neon_inner.o
Signature · 1 line
2.20.1
author maintainer
#28 lore raw

PATCH 29/78 KVM: s390: vsie: Fix possible race when shadowing region 3 tables

In reply to Sasha Levin

From: David Hildenbrand <[email protected]>

[ Upstream commit 1493e0f944f3c319d11e067c185c904d01c17ae5 ]

We have to properly retry again by returning -EINVAL immediately in case somebody else instantiated the table concurrently. We missed to add the goto in this function only. The code now matches the other, similar shadowing functions.

We are overwriting an existing region 2 table entry. All allocated pages are added to the crst_list to be freed later, so they are not lost forever. However, when unshadowing the region 2 table, we wouldn't trigger unshadowing of the original shadowed region 3 table that we replaced. It would get unshadowed when the original region 3 table is modified. As it's not connected to the page table hierarchy anymore, it's not going to get used anymore. However, for a limited time, this page table will stick around, so it's in some sense a temporary memory leak.

Identified by manual code inspection. I don't think this classifies as stable material.

---
 arch/s390/mm/gmap.c | 1 +
 1 file changed, 1 insertion(+)
arch/s390/mm/gmap.c +1 -0
index edcdca97e85ee..06d602c5ec7b7 100644
@@ -1840,6 +1840,7 @@ int gmap_shadow_r3t(struct gmap *sg, unsigned long saddr, unsigned long r3t,
 		goto out_free;
 	} else if (*table & _REGION_ENTRY_ORIGIN) {
 		rc = -EAGAIN;		/* Race with shadow */
+		goto out_free;
 	}
 	crst_table_init(s_r3t, _REGION3_ENTRY_EMPTY);
 	/* mark as invalid as long as the parent table is not protected */
Signature · 1 line
2.20.1
author maintainer
#29 lore raw

PATCH 30/78 drm/nouveau: workaround runpm fail by disabling PCI power management on certain intel bridges

In reply to Sasha Levin

From: Karol Herbst <[email protected]>

[ Upstream commit 434fdb51513bf3057ac144d152e6f2f2b509e857 ]

Fixes the infamous 'runtime PM' bug many users are facing on Laptops with Nvidia Pascal GPUs by skipping said PCI power state changes on the GPU.

Depending on the used kernel there might be messages like those in demsg:

"nouveau 0000:01:00.0: Refused to change power state, currently in D3" "nouveau 0000:01:00.0: can't change power state from D3cold to D0 (config space inaccessible)"
followed by backtraces of kernel crashes or timeouts within nouveau.

It's still unkown why this issue exists, but this is a reliable workaround and solves a very annoying issue for user having to choose between a crashing kernel or higher power consumption of their Laptops.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=205623

---
 drivers/gpu/drm/nouveau/nouveau_drm.c | 63 +++++++++++++++++++++++++++
 drivers/gpu/drm/nouveau/nouveau_drv.h |  2 +
 2 files changed, 65 insertions(+)
drivers/gpu/drm/nouveau/nouveau_drm.c +63 -0
index 2cd83849600f3..b1beed40e746a 100644
@@ -618,6 +618,64 @@ nouveau_drm_device_fini(struct drm_device *dev)
 	kfree(drm);
 }
 
+/*
+ * On some Intel PCIe bridge controllers doing a
+ * D0 -> D3hot -> D3cold -> D0 sequence causes Nvidia GPUs to not reappear.
+ * Skipping the intermediate D3hot step seems to make it work again. This is
+ * probably caused by not meeting the expectation the involved AML code has
+ * when the GPU is put into D3hot state before invoking it.
+ *
+ * This leads to various manifestations of this issue:
+ *  - AML code execution to power on the GPU hits an infinite loop (as the
+ *    code waits on device memory to change).
+ *  - kernel crashes, as all PCI reads return -1, which most code isn't able
+ *    to handle well enough.
+ *
+ * In all cases dmesg will contain at least one line like this:
+ * 'nouveau 0000:01:00.0: Refused to change power state, currently in D3'
+ * followed by a lot of nouveau timeouts.
+ *
+ * In the \_SB.PCI0.PEG0.PG00._OFF code deeper down writes bit 0x80 to the not
+ * documented PCI config space register 0x248 of the Intel PCIe bridge
+ * controller (0x1901) in order to change the state of the PCIe link between
+ * the PCIe port and the GPU. There are alternative code paths using other
+ * registers, which seem to work fine (executed pre Windows 8):
+ *  - 0xbc bit 0x20 (publicly available documentation claims 'reserved')
+ *  - 0xb0 bit 0x10 (link disable)
+ * Changing the conditions inside the firmware by poking into the relevant
+ * addresses does resolve the issue, but it seemed to be ACPI private memory
+ * and not any device accessible memory at all, so there is no portable way of
+ * changing the conditions.
+ * On a XPS 9560 that means bits [0,3] on \CPEX need to be cleared.
+ *
+ * The only systems where this behavior can be seen are hybrid graphics laptops
+ * with a secondary Nvidia Maxwell, Pascal or Turing GPU. It's unclear whether
+ * this issue only occurs in combination with listed Intel PCIe bridge
+ * controllers and the mentioned GPUs or other devices as well.
+ *
+ * documentation on the PCIe bridge controller can be found in the
+ * "7th Generation Intel® Processor Families for H Platforms Datasheet Volume 2"
+ * Section "12 PCI Express* Controller (x16) Registers"
+ */
+
+static void quirk_broken_nv_runpm(struct pci_dev *pdev)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct nouveau_drm *drm = nouveau_drm(dev);
+	struct pci_dev *bridge = pci_upstream_bridge(pdev);
+
+	if (!bridge || bridge->vendor != PCI_VENDOR_ID_INTEL)
+		return;
+
+	switch (bridge->device) {
+	case 0x1901:
+		drm->old_pm_cap = pdev->pm_cap;
+		pdev->pm_cap = 0;
+		NV_INFO(drm, "Disabling PCI power management to avoid bug\n");
+		break;
+	}
+}
+
 static int nouveau_drm_probe(struct pci_dev *pdev,
 			     const struct pci_device_id *pent)
 {
@@ -699,6 +757,7 @@ static int nouveau_drm_probe(struct pci_dev *pdev,
 	if (ret)
 		goto fail_drm_dev_init;
 
+	quirk_broken_nv_runpm(pdev);
 	return 0;
 
 fail_drm_dev_init:
@@ -736,7 +795,11 @@ static void
 nouveau_drm_remove(struct pci_dev *pdev)
 {
 	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct nouveau_drm *drm = nouveau_drm(dev);
 
+	/* revert our workaround */
+	if (drm->old_pm_cap)
+		pdev->pm_cap = drm->old_pm_cap;
 	nouveau_drm_device_remove(dev);
 }
 
drivers/gpu/drm/nouveau/nouveau_drv.h +2 -0
index 70f34cacc552c..8104e3806499d 100644
@@ -138,6 +138,8 @@ struct nouveau_drm {
 
 	struct list_head clients;
 
+	u8 old_pm_cap;
+
 	struct {
 		struct agp_bridge_data *bridge;
 		u32 base;
Signature · 6 lines
2.20.1

_______________________________________________
Nouveau mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/nouveau
author maintainer
#30 lore raw

PATCH 31/78 s390/cio: generate delayed uevent for vfio-ccw subchannels

In reply to Sasha Levin

From: Cornelia Huck <[email protected]>

[ Upstream commit 2bc55eaeb88d30accfc1b6ac2708d4e4b81ca260 ]

The common I/O layer delays the ADD uevent for subchannels and delegates generating this uevent to the individual subchannel drivers. The vfio-ccw I/O subchannel driver, however, did not do that, and will not generate an ADD uevent for subchannels that had not been bound to a different driver (or none at all, which also triggers the uevent).

Generate the ADD uevent at the end of the probe function if uevents were still suppressed for the device.

Message-Id: <[email protected]>

Fixes 63f1934d562d ("vfio: ccw: basic implementation for vfio_ccw driver")
---
 drivers/s390/cio/vfio_ccw_drv.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/s390/cio/vfio_ccw_drv.c +5 -0
index e401a3d0aa570..339a6bc0339b0 100644
@@ -167,6 +167,11 @@ static int vfio_ccw_sch_probe(struct subchannel *sch)
 	if (ret)
 		goto out_disable;
 
+	if (dev_get_uevent_suppress(&sch->dev)) {
+		dev_set_uevent_suppress(&sch->dev, 0);
+		kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+	}
+
 	VFIO_CCW_MSG_EVENT(4, "bound to subchannel %x.%x.%04x\n",
 			   sch->schid.cssid, sch->schid.ssid,
 			   sch->schid.sch_no);
Signature · 1 line
2.20.1
author maintainer
#31 lore raw

PATCH 32/78 s390/cio: avoid duplicated 'ADD' uevents

In reply to Sasha Levin

From: Cornelia Huck <[email protected]>

[ Upstream commit 05ce3e53f375295c2940390b2b429e506e07655c ]

The common I/O layer delays the ADD uevent for subchannels and delegates generating this uevent to the individual subchannel drivers. The io_subchannel driver will do so when the associated ccw_device has been registered -- but unconditionally, so more ADD uevents will be generated if a subchannel has been unbound from the io_subchannel driver and later rebound.

To fix this, only generate the ADD event if uevents were still suppressed for the device.

Fixes fa1a8c23eb7d ("s390: cio: Delay uevents for subchannels")

Message-Id: <[email protected]>

---
 drivers/s390/cio/device.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)
drivers/s390/cio/device.c +9 -4
index 0c6245fc77069..983f9c9e08deb 100644
@@ -849,8 +849,10 @@ static void io_subchannel_register(struct ccw_device *cdev)
 	 * Now we know this subchannel will stay, we can throw
 	 * our delayed uevent.
 	 */
-	dev_set_uevent_suppress(&sch->dev, 0);
-	kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+	if (dev_get_uevent_suppress(&sch->dev)) {
+		dev_set_uevent_suppress(&sch->dev, 0);
+		kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+	}
 	/* make it known to the system */
 	ret = ccw_device_add(cdev);
 	if (ret) {
@@ -1058,8 +1060,11 @@ static int io_subchannel_probe(struct subchannel *sch)
 		 * Throw the delayed uevent for the subchannel, register
 		 * the ccw_device and exit.
 		 */
-		dev_set_uevent_suppress(&sch->dev, 0);
-		kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+		if (dev_get_uevent_suppress(&sch->dev)) {
+			/* should always be the case for the console */
+			dev_set_uevent_suppress(&sch->dev, 0);
+			kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+		}
 		cdev = sch_get_cdev(sch);
 		rc = ccw_device_add(cdev);
 		if (rc) {
Signature · 1 line
2.20.1
author maintainer
#32 lore raw

PATCH 33/78 loop: Better discard support for block devices

In reply to Sasha Levin

From: Evan Green <[email protected]>

[ Upstream commit c52abf563049e787c1341cdf15c7dbe1bfbc951b ]

If the backing device for a loop device is itself a block device, then mirror the "write zeroes" capabilities of the underlying block device into the loop device. Copy this capability into both max_write_zeroes_sectors and max_discard_sectors of the loop device.

The reason for this is that REQ_OP_DISCARD on a loop device translates into blkdev_issue_zeroout(), rather than blkdev_issue_discard(). This presents a consistent interface for loop devices (that discarded data is zeroed), regardless of the backing device type of the loop device. There should be no behavior change for loop devices backed by regular files.

This change fixes blktest block/003, and removes an extraneous error print in block/013 when testing on a loop device backed by a block device that does not support discard.

[used updated version of Evan's comment in loop_config_discard()]
[moved backingq to local scope, removed redundant braces]
---
 drivers/block/loop.c | 42 +++++++++++++++++++++++++++++++-----------
 1 file changed, 31 insertions(+), 11 deletions(-)
drivers/block/loop.c +31 -11
index ef6e251857c8c..57ed6b70d2950 100644
@@ -427,11 +427,12 @@ static int lo_fallocate(struct loop_device *lo, struct request *rq, loff_t pos,
 	 * information.
 	 */
 	struct file *file = lo->lo_backing_file;
+	struct request_queue *q = lo->lo_queue;
 	int ret;
 
 	mode |= FALLOC_FL_KEEP_SIZE;
 
-	if ((!file->f_op->fallocate) || lo->lo_encrypt_key_size) {
+	if (!blk_queue_discard(q)) {
 		ret = -EOPNOTSUPP;
 		goto out;
 	}
@@ -863,28 +864,47 @@ static void loop_config_discard(struct loop_device *lo)
 	struct inode *inode = file->f_mapping->host;
 	struct request_queue *q = lo->lo_queue;
 
+	/*
+	 * If the backing device is a block device, mirror its zeroing
+	 * capability. Set the discard sectors to the block device's zeroing
+	 * capabilities because loop discards result in blkdev_issue_zeroout(),
+	 * not blkdev_issue_discard(). This maintains consistent behavior with
+	 * file-backed loop devices: discarded regions read back as zero.
+	 */
+	if (S_ISBLK(inode->i_mode) && !lo->lo_encrypt_key_size) {
+		struct request_queue *backingq;
+
+		backingq = bdev_get_queue(inode->i_bdev);
+		blk_queue_max_discard_sectors(q,
+			backingq->limits.max_write_zeroes_sectors);
+
+		blk_queue_max_write_zeroes_sectors(q,
+			backingq->limits.max_write_zeroes_sectors);
+
 	/*
 	 * We use punch hole to reclaim the free space used by the
 	 * image a.k.a. discard. However we do not support discard if
 	 * encryption is enabled, because it may give an attacker
 	 * useful information.
 	 */
-	if ((!file->f_op->fallocate) ||
-	    lo->lo_encrypt_key_size) {
+	} else if (!file->f_op->fallocate || lo->lo_encrypt_key_size) {
 		q->limits.discard_granularity = 0;
 		q->limits.discard_alignment = 0;
 		blk_queue_max_discard_sectors(q, 0);
 		blk_queue_max_write_zeroes_sectors(q, 0);
-		blk_queue_flag_clear(QUEUE_FLAG_DISCARD, q);
-		return;
-	}
 
-	q->limits.discard_granularity = inode->i_sb->s_blocksize;
-	q->limits.discard_alignment = 0;
+	} else {
+		q->limits.discard_granularity = inode->i_sb->s_blocksize;
+		q->limits.discard_alignment = 0;
 
-	blk_queue_max_discard_sectors(q, UINT_MAX >> 9);
-	blk_queue_max_write_zeroes_sectors(q, UINT_MAX >> 9);
-	blk_queue_flag_set(QUEUE_FLAG_DISCARD, q);
+		blk_queue_max_discard_sectors(q, UINT_MAX >> 9);
+		blk_queue_max_write_zeroes_sectors(q, UINT_MAX >> 9);
+	}
+
+	if (q->limits.max_write_zeroes_sectors)
+		blk_queue_flag_set(QUEUE_FLAG_DISCARD, q);
+	else
+		blk_queue_flag_clear(QUEUE_FLAG_DISCARD, q);
 }
 
 static void loop_unprepare_queue(struct loop_device *lo)
Signature · 1 line
2.20.1
author maintainer
#33 lore raw

PATCH 34/78 pwm: pca9685: Fix PWM/GPIO inter-operation

In reply to Sasha Levin

From: Sven Van Asbroeck <[email protected]>

[ Upstream commit 9cc5f232a4b6a0ef6e9b57876d61b88f61bdd7c2 ]

This driver allows pwms to be requested as gpios via gpiolib. Obviously, it should not be allowed to request a GPIO when its corresponding PWM is already requested (and vice versa). So it requires some exclusion code.

Given that the PWMm and GPIO cores are not synchronized with respect to each other, this exclusion code will also require proper synchronization.

Such a mechanism was in place, but was inadvertently removed by Uwe's clean-up in commit e926b12c611c ("pwm: Clear chip_data in pwm_put()").

Upon revisiting the synchronization mechanism, we found that theoretically, it could allow two threads to successfully request conflicting PWMs/GPIOs.

Replace with a bitmap which tracks PWMs in-use, plus a mutex. As long as PWM and GPIO's respective request/free functions modify the in-use bitmap while holding the mutex, proper synchronization will be guaranteed.

---
 drivers/pwm/pwm-pca9685.c | 85 ++++++++++++++++++++++-----------------
 1 file changed, 48 insertions(+), 37 deletions(-)
drivers/pwm/pwm-pca9685.c +48 -37
index b07bdca3d510d..590375be52147 100644
@@ -20,6 +20,7 @@
 #include <linux/slab.h>
 #include <linux/delay.h>
 #include <linux/pm_runtime.h>
+#include <linux/bitmap.h>
 
 /*
  * Because the PCA9685 has only one prescaler per chip, changing the period of
@@ -74,6 +75,7 @@ struct pca9685 {
 #if IS_ENABLED(CONFIG_GPIOLIB)
 	struct mutex lock;
 	struct gpio_chip gpio;
+	DECLARE_BITMAP(pwms_inuse, PCA9685_MAXCHAN + 1);
 #endif
 };
 
@@ -83,51 +85,51 @@ static inline struct pca9685 *to_pca(struct pwm_chip *chip)
 }
 
 #if IS_ENABLED(CONFIG_GPIOLIB)
-static int pca9685_pwm_gpio_request(struct gpio_chip *gpio, unsigned int offset)
+static bool pca9685_pwm_test_and_set_inuse(struct pca9685 *pca, int pwm_idx)
 {
-	struct pca9685 *pca = gpiochip_get_data(gpio);
-	struct pwm_device *pwm;
+	bool is_inuse;
 
 	mutex_lock(&pca->lock);
-
-	pwm = &pca->chip.pwms[offset];
-
-	if (pwm->flags & (PWMF_REQUESTED | PWMF_EXPORTED)) {
-		mutex_unlock(&pca->lock);
-		return -EBUSY;
+	if (pwm_idx >= PCA9685_MAXCHAN) {
+		/*
+		 * "all LEDs" channel:
+		 * pretend already in use if any of the PWMs are requested
+		 */
+		if (!bitmap_empty(pca->pwms_inuse, PCA9685_MAXCHAN)) {
+			is_inuse = true;
+			goto out;
+		}
+	} else {
+		/*
+		 * regular channel:
+		 * pretend already in use if the "all LEDs" channel is requested
+		 */
+		if (test_bit(PCA9685_MAXCHAN, pca->pwms_inuse)) {
+			is_inuse = true;
+			goto out;
+		}
 	}
-
-	pwm_set_chip_data(pwm, (void *)1);
-
+	is_inuse = test_and_set_bit(pwm_idx, pca->pwms_inuse);
+out:
 	mutex_unlock(&pca->lock);
-	pm_runtime_get_sync(pca->chip.dev);
-	return 0;
+	return is_inuse;
 }
 
-static bool pca9685_pwm_is_gpio(struct pca9685 *pca, struct pwm_device *pwm)
+static void pca9685_pwm_clear_inuse(struct pca9685 *pca, int pwm_idx)
 {
-	bool is_gpio = false;
-
 	mutex_lock(&pca->lock);
+	clear_bit(pwm_idx, pca->pwms_inuse);
+	mutex_unlock(&pca->lock);
+}
 
-	if (pwm->hwpwm >= PCA9685_MAXCHAN) {
-		unsigned int i;
-
-		/*
-		 * Check if any of the GPIOs are requested and in that case
-		 * prevent using the "all LEDs" channel.
-		 */
-		for (i = 0; i < pca->gpio.ngpio; i++)
-			if (gpiochip_is_requested(&pca->gpio, i)) {
-				is_gpio = true;
-				break;
-			}
-	} else if (pwm_get_chip_data(pwm)) {
-		is_gpio = true;
-	}
+static int pca9685_pwm_gpio_request(struct gpio_chip *gpio, unsigned int offset)
+{
+	struct pca9685 *pca = gpiochip_get_data(gpio);
 
-	mutex_unlock(&pca->lock);
-	return is_gpio;
+	if (pca9685_pwm_test_and_set_inuse(pca, offset))
+		return -EBUSY;
+	pm_runtime_get_sync(pca->chip.dev);
+	return 0;
 }
 
 static int pca9685_pwm_gpio_get(struct gpio_chip *gpio, unsigned int offset)
@@ -162,6 +164,7 @@ static void pca9685_pwm_gpio_free(struct gpio_chip *gpio, unsigned int offset)
 
 	pca9685_pwm_gpio_set(gpio, offset, 0);
 	pm_runtime_put(pca->chip.dev);
+	pca9685_pwm_clear_inuse(pca, offset);
 }
 
 static int pca9685_pwm_gpio_get_direction(struct gpio_chip *chip,
@@ -213,12 +216,17 @@ static int pca9685_pwm_gpio_probe(struct pca9685 *pca)
 	return devm_gpiochip_add_data(dev, &pca->gpio, pca);
 }
 #else
-static inline bool pca9685_pwm_is_gpio(struct pca9685 *pca,
-				       struct pwm_device *pwm)
+static inline bool pca9685_pwm_test_and_set_inuse(struct pca9685 *pca,
+						  int pwm_idx)
 {
 	return false;
 }
 
+static inline void
+pca9685_pwm_clear_inuse(struct pca9685 *pca, int pwm_idx)
+{
+}
+
 static inline int pca9685_pwm_gpio_probe(struct pca9685 *pca)
 {
 	return 0;
@@ -402,7 +410,7 @@ static int pca9685_pwm_request(struct pwm_chip *chip, struct pwm_device *pwm)
 {
 	struct pca9685 *pca = to_pca(chip);
 
-	if (pca9685_pwm_is_gpio(pca, pwm))
+	if (pca9685_pwm_test_and_set_inuse(pca, pwm->hwpwm))
 		return -EBUSY;
 	pm_runtime_get_sync(chip->dev);
 
@@ -411,8 +419,11 @@ static int pca9685_pwm_request(struct pwm_chip *chip, struct pwm_device *pwm)
 
 static void pca9685_pwm_free(struct pwm_chip *chip, struct pwm_device *pwm)
 {
+	struct pca9685 *pca = to_pca(chip);
+
 	pca9685_pwm_disable(chip, pwm);
 	pm_runtime_put(chip->dev);
+	pca9685_pwm_clear_inuse(pca, pwm->hwpwm);
 }
 
 static const struct pwm_ops pca9685_pwm_ops = {
Signature · 1 line
2.20.1
author maintainer
#34 lore raw

PATCH 35/78 Revert "powerpc/64: irq_work avoid interrupt when called with hardware irqs enabled"

In reply to Sasha Levin

From: Nicholas Piggin <[email protected]>

[ Upstream commit abc3fce76adbdfa8f87272c784b388cd20b46049 ]

This reverts commit ebb37cf3ffd39fdb6ec5b07111f8bb2f11d92c5f.

That commit does not play well with soft-masked irq state manipulations in idle, interrupt replay, and possibly others due to tracing code sometimes using irq_work_queue (e.g., in trace_hardirqs_on()). That can cause PACA_IRQ_DEC to become set when it is not expected, and be ignored or cleared or cause warnings.

The net result seems to be missing an irq_work until the next timer interrupt in the worst case which is usually not going to be noticed, however it could be a long time if the tick is disabled, which is against the spirit of irq_work and might cause real problems.

The idea is still solid, but it would need more work. It's not really clear if it would be worth added complexity, so revert this for now (not a straight revert, but replace with a comment explaining why we might see interrupts happening, and gives git blame something to find).

Fixes ebb37cf3ffd3 ("powerpc/64: irq_work avoid interrupt when called with hardware irqs enabled")
---
 arch/powerpc/kernel/time.c | 44 +++++++++++---------------------------
 1 file changed, 13 insertions(+), 31 deletions(-)
arch/powerpc/kernel/time.c +13 -31
index 11301a1187f33..0e0a2227af7d7 100644
@@ -522,35 +522,6 @@ static inline void clear_irq_work_pending(void)
 		"i" (offsetof(struct paca_struct, irq_work_pending)));
 }
 
-void arch_irq_work_raise(void)
-{
-	preempt_disable();
-	set_irq_work_pending_flag();
-	/*
-	 * Non-nmi code running with interrupts disabled will replay
-	 * irq_happened before it re-enables interrupts, so setthe
-	 * decrementer there instead of causing a hardware exception
-	 * which would immediately hit the masked interrupt handler
-	 * and have the net effect of setting the decrementer in
-	 * irq_happened.
-	 *
-	 * NMI interrupts can not check this when they return, so the
-	 * decrementer hardware exception is raised, which will fire
-	 * when interrupts are next enabled.
-	 *
-	 * BookE does not support this yet, it must audit all NMI
-	 * interrupt handlers to ensure they call nmi_enter() so this
-	 * check would be correct.
-	 */
-	if (IS_ENABLED(CONFIG_BOOKE) || !irqs_disabled() || in_nmi()) {
-		set_dec(1);
-	} else {
-		hard_irq_disable();
-		local_paca->irq_happened |= PACA_IRQ_DEC;
-	}
-	preempt_enable();
-}
-
 #else /* 32-bit */
 
 DEFINE_PER_CPU(u8, irq_work_pending);
@@ -559,16 +530,27 @@ DEFINE_PER_CPU(u8, irq_work_pending);
 #define test_irq_work_pending()		__this_cpu_read(irq_work_pending)
 #define clear_irq_work_pending()	__this_cpu_write(irq_work_pending, 0)
 
+#endif /* 32 vs 64 bit */
+
 void arch_irq_work_raise(void)
 {
+	/*
+	 * 64-bit code that uses irq soft-mask can just cause an immediate
+	 * interrupt here that gets soft masked, if this is called under
+	 * local_irq_disable(). It might be possible to prevent that happening
+	 * by noticing interrupts are disabled and setting decrementer pending
+	 * to be replayed when irqs are enabled. The problem there is that
+	 * tracing can call irq_work_raise, including in code that does low
+	 * level manipulations of irq soft-mask state (e.g., trace_hardirqs_on)
+	 * which could get tangled up if we're messing with the same state
+	 * here.
+	 */
 	preempt_disable();
 	set_irq_work_pending_flag();
 	set_dec(1);
 	preempt_enable();
 }
 
-#endif /* 32 vs 64 bit */
-
 #else  /* CONFIG_IRQ_WORK */
 
 #define test_irq_work_pending()	0
Signature · 1 line
2.20.1
author maintainer
#35 lore raw

PATCH 36/78 powerpc/pseries: Fix MCE handling on pseries

In reply to Sasha Levin

From: Ganesh Goudar <[email protected]>

[ Upstream commit a95a0a1654f16366360399574e10efd87e867b39 ]

MCE handling on pSeries platform fails as recent rework to use common code for pSeries and PowerNV in machine check error handling tries to access per-cpu variables in realmode. The per-cpu variables may be outside the RMO region on pSeries platform and needs translation to be enabled for access. Just moving these per-cpu variable into RMO region did'nt help because we queue some work to workqueues in real mode, which again tries to touch per-cpu variables. Also fwnmi_release_errinfo() cannot be called when translation is not enabled.

This patch fixes this by enabling translation in the exception handler when all required real mode handling is done. This change only affects the pSeries platform.

Without this fix below kernel crash is seen on injecting SLB multihit:

BUG: Unable to handle kernel data access on read at 0xc00000027b205950
Faulting instruction address: 0xc00000000003b7e0
Oops: Kernel access of bad area, sig: 11 [#1]
LE PAGE_SIZE=64K MMU=Hash SMP NR_CPUS=2048 NUMA pSeries
Modules linked in: mcetest_slb(OE+) af_packet(E) xt_tcpudp(E) ip6t_rpfilter(E) ip6t_REJECT(E) ipt_REJECT(E) xt_conntrack(E) ip_set(E) nfnetlink(E) ebtable_nat(E) ebtable_broute(E) ip6table_nat(E) ip6table_mangle(E) ip6table_raw(E) ip6table_security(E) iptable_nat(E) nf_nat(E) nf_conntrack(E) nf_defrag_ipv6(E) nf_defrag_ipv4(E) iptable_mangle(E) iptable_raw(E) iptable_security(E) ebtable_filter(E) ebtables(E) ip6table_filter(E) ip6_tables(E) iptable_filter(E) ip_tables(E) x_tables(E) xfs(E) ibmveth(E) vmx_crypto(E) gf128mul(E) uio_pdrv_genirq(E) uio(E) crct10dif_vpmsum(E) rtc_generic(E) btrfs(E) libcrc32c(E) xor(E) zstd_decompress(E) zstd_compress(E) raid6_pq(E) sr_mod(E) sd_mod(E) cdrom(E) ibmvscsi(E) scsi_transport_srp(E) crc32c_vpmsum(E) dm_mod(E) sg(E) scsi_mod(E)
CPU: 34 PID: 8154 Comm: insmod Kdump: loaded Tainted: G OE 5.5.0-mahesh #1
NIP: c00000000003b7e0 LR: c0000000000f2218 CTR: 0000000000000000
REGS: c000000007dcb960 TRAP: 0300 Tainted: G OE (5.5.0-mahesh)
MSR: 8000000000001003 <SF,ME,RI,LE> CR: 28002428 XER: 20040000
CFAR: c0000000000f2214 DAR: c00000027b205950 DSISR: 40000000 IRQMASK: 0

GPR00: c0000000000f2218 c000000007dcbbf0 c000000001544800 c000000007dcbd70
GPR04: 0000000000000001 c000000007dcbc98 c008000000d00258 c0080000011c0000
GPR08: 0000000000000000 0000000300000003 c000000001035950 0000000003000048
GPR12: 000000027a1d0000 c000000007f9c000 0000000000000558 0000000000000000
GPR16: 0000000000000540 c008000001110000 c008000001110540 0000000000000000
GPR20: c00000000022af10 c00000025480fd70 c008000001280000 c00000004bfbb300
GPR24: c000000001442330 c00800000800000d c008000008000000 4009287a77000510
GPR28: 0000000000000000 0000000000000002 c000000001033d30 0000000000000001 NIP [c00000000003b7e0] save_mce_event+0x30/0x240
LR [c0000000000f2218] pseries_machine_check_realmode+0x2c8/0x4f0

Call Trace:
Instruction dump:
3c4c0151 38429050 7c0802a6 60000000 fbc1fff0 fbe1fff8 f821ffd1 3d42ffaf
3fc2ffaf e98d0030 394a1150 3bdef530 <7d6a62aa> 1d2b0048 2f8b0063 380b0001
---[ end trace 46fd63f36bbdd940 ]---
Fixes 9ca766f9891d ("powerpc/64s/pseries: machine check convert to use common event code")
---
 arch/powerpc/platforms/pseries/ras.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
arch/powerpc/platforms/pseries/ras.c +11 -0
index 3acdcc3bb908c..753adeb624f23 100644
@@ -683,6 +683,17 @@ static int mce_handle_error(struct pt_regs *regs, struct rtas_error_log *errp)
 #endif
 
 out:
+	/*
+	 * Enable translation as we will be accessing per-cpu variables
+	 * in save_mce_event() which may fall outside RMO region, also
+	 * leave it enabled because subsequently we will be queuing work
+	 * to workqueues where again per-cpu variables accessed, besides
+	 * fwnmi_release_errinfo() crashes when called in realmode on
+	 * pseries.
+	 * Note: All the realmode handling like flushing SLB entries for
+	 *       SLB multihit is done by now.
+	 */
+	mtmsr(mfmsr() | MSR_IR | MSR_DR);
 	save_mce_event(regs, disposition == RTAS_DISP_FULLY_RECOVERED,
 			&mce_err, regs->nip, eaddr, paddr);
 
Signature · 1 line
2.20.1
author maintainer
#36 lore raw

PATCH 37/78 drm/amdkfd: kfree the wrong pointer

In reply to Sasha Levin

From: Jack Zhang <[email protected]>

[ Upstream commit 3148a6a0ef3cf93570f30a477292768f7eb5d3c3 ]

Originally, it kfrees the wrong pointer for mem_obj. It would cause memory leak under stress test.

---
 drivers/gpu/drm/amd/amdkfd/kfd_device.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/amdkfd/kfd_device.c +2 -2
index 0dc1084b5e829..ad9483b9eea32 100644
@@ -1112,9 +1112,9 @@ int kfd_gtt_sa_allocate(struct kfd_dev *kfd, unsigned int size,
 	return 0;
 
 kfd_gtt_no_free_chunk:
-	pr_debug("Allocation failed with mem_obj = %p\n", mem_obj);
+	pr_debug("Allocation failed with mem_obj = %p\n", *mem_obj);
 	mutex_unlock(&kfd->gtt_sa_lock);
-	kfree(mem_obj);
+	kfree(*mem_obj);
 	return -ENOMEM;
 }
 
Signature · 6 lines
2.20.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author maintainer
#37 lore raw

PATCH 38/78 nvme: fix compat address handling in several ioctls

In reply to Sasha Levin

From: Nick Bowler <[email protected]>

[ Upstream commit c95b708d5fa65b4e51f088ee077d127fd5a57b70 ]

On a 32-bit kernel, the upper bits of userspace addresses passed via various ioctls are silently ignored by the nvme driver.

However on a 64-bit kernel running a compat task, these upper bits are not ignored and are in fact required to be zero for the ioctls to work.

Unfortunately, this difference matters. 32-bit smartctl submits the NVME_IOCTL_ADMIN_CMD ioctl with garbage in these upper bits because it seems the pointer value it puts into the nvme_passthru_cmd structure is sign extended. This works fine on 32-bit kernels but fails on a 64-bit one because (at least on my setup) the addresses smartctl uses are consistently above 2G. For example:

  # smartctl -x /dev/nvme0n1
  smartctl 7.1 2019-12-30 r5022 [x86_64-linux-5.5.11] (local build)
  Copyright (C) 2002-19, Bruce Allen, Christian Franke, www.smartmontools.org

Read NVMe Identify Controller failed: NVME_IOCTL_ADMIN_CMD: Bad address

Since changing 32-bit kernels to actually check all of the submitted address bits now would break existing userspace, this patch fixes the compat problem by explicitly zeroing the upper bits in the compat case. This enables 32-bit smartctl to work on a 64-bit kernel.

---
 drivers/nvme/host/core.c | 27 ++++++++++++++++++++-------
 1 file changed, 20 insertions(+), 7 deletions(-)
drivers/nvme/host/core.c +20 -7
index b8fe42f4b3c5b..f97c48fd3edae 100644
@@ -6,6 +6,7 @@
 
 #include <linux/blkdev.h>
 #include <linux/blk-mq.h>
+#include <linux/compat.h>
 #include <linux/delay.h>
 #include <linux/errno.h>
 #include <linux/hdreg.h>
@@ -1244,6 +1245,18 @@ static void nvme_enable_aen(struct nvme_ctrl *ctrl)
 	queue_work(nvme_wq, &ctrl->async_event_work);
 }
 
+/*
+ * Convert integer values from ioctl structures to user pointers, silently
+ * ignoring the upper bits in the compat case to match behaviour of 32-bit
+ * kernels.
+ */
+static void __user *nvme_to_user_ptr(uintptr_t ptrval)
+{
+	if (in_compat_syscall())
+		ptrval = (compat_uptr_t)ptrval;
+	return (void __user *)ptrval;
+}
+
 static int nvme_submit_io(struct nvme_ns *ns, struct nvme_user_io __user *uio)
 {
 	struct nvme_user_io io;
@@ -1267,7 +1280,7 @@ static int nvme_submit_io(struct nvme_ns *ns, struct nvme_user_io __user *uio)
 
 	length = (io.nblocks + 1) << ns->lba_shift;
 	meta_len = (io.nblocks + 1) * ns->ms;
-	metadata = (void __user *)(uintptr_t)io.metadata;
+	metadata = nvme_to_user_ptr(io.metadata);
 
 	if (ns->ext) {
 		length += meta_len;
@@ -1290,7 +1303,7 @@ static int nvme_submit_io(struct nvme_ns *ns, struct nvme_user_io __user *uio)
 	c.rw.appmask = cpu_to_le16(io.appmask);
 
 	return nvme_submit_user_cmd(ns->queue, &c,
-			(void __user *)(uintptr_t)io.addr, length,
+			nvme_to_user_ptr(io.addr), length,
 			metadata, meta_len, lower_32_bits(io.slba), NULL, 0);
 }
 
@@ -1410,9 +1423,9 @@ static int nvme_user_cmd(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
 
 	effects = nvme_passthru_start(ctrl, ns, cmd.opcode);
 	status = nvme_submit_user_cmd(ns ? ns->queue : ctrl->admin_q, &c,
-			(void __user *)(uintptr_t)cmd.addr, cmd.data_len,
-			(void __user *)(uintptr_t)cmd.metadata,
-			cmd.metadata_len, 0, &result, timeout);
+			nvme_to_user_ptr(cmd.addr), cmd.data_len,
+			nvme_to_user_ptr(cmd.metadata), cmd.metadata_len,
+			0, &result, timeout);
 	nvme_passthru_end(ctrl, effects);
 
 	if (status >= 0) {
@@ -1457,8 +1470,8 @@ static int nvme_user_cmd64(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
 
 	effects = nvme_passthru_start(ctrl, ns, cmd.opcode);
 	status = nvme_submit_user_cmd(ns ? ns->queue : ctrl->admin_q, &c,
-			(void __user *)(uintptr_t)cmd.addr, cmd.data_len,
-			(void __user *)(uintptr_t)cmd.metadata, cmd.metadata_len,
+			nvme_to_user_ptr(cmd.addr), cmd.data_len,
+			nvme_to_user_ptr(cmd.metadata), cmd.metadata_len,
 			0, &cmd.result, timeout);
 	nvme_passthru_end(ctrl, effects);
 
Signature · 7 lines
2.20.1


_______________________________________________
linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme
#38 lore raw

Re: [PATCH AUTOSEL 5.4 38/78] nvme: fix compat address handling in several ioctls

Quoted 51 lines · Sasha Levin

From: Nick Bowler <[email protected]>

[ Upstream commit c95b708d5fa65b4e51f088ee077d127fd5a57b70 ]

On a 32-bit kernel, the upper bits of userspace addresses passed via various ioctls are silently ignored by the nvme driver.

However on a 64-bit kernel running a compat task, these upper bits are not ignored and are in fact required to be zero for the ioctls to work.

Unfortunately, this difference matters. 32-bit smartctl submits the NVME_IOCTL_ADMIN_CMD ioctl with garbage in these upper bits because it seems the pointer value it puts into the nvme_passthru_cmd structure is sign extended. This works fine on 32-bit kernels but fails on a 64-bit one because (at least on my setup) the addresses smartctl uses are consistently above 2G. For example:

  # smartctl -x /dev/nvme0n1
  smartctl 7.1 2019-12-30 r5022 [x86_64-linux-5.5.11] (local build)
  Copyright (C) 2002-19, Bruce Allen, Christian Franke, www.smartmontools.org

Read NVMe Identify Controller failed: NVME_IOCTL_ADMIN_CMD: Bad address

Since changing 32-bit kernels to actually check all of the submitted address bits now would break existing userspace, this patch fixes the compat problem by explicitly zeroing the upper bits in the compat case. This enables 32-bit smartctl to work on a 64-bit kernel.

---
 drivers/nvme/host/core.c | 27 ++++++++++++++++++++-------
 1 file changed, 20 insertions(+), 7 deletions(-)
drivers/nvme/host/core.c +1 -0
index b8fe42f4b3c5b..f97c48fd3edae 100644
@@ -6,6 +6,7 @@
 
 #include <linux/blkdev.h>
 #include <linux/blk-mq.h>
+#include <linux/compat.h>
 #include <linux/delay.h>
 #include <linux/errno.h>
 #include <linux/hdreg.h>
@@ -1244,6 +1245,18 @@ static void nvme_enable_aen(struct nvme_ctrl *ctrl)
        queue_work(nvme_wq, &ctrl->async_event_work);
 }

drivers/nvme/host/core.c static void nvme_enable_aen(struct nvme_ctrl *ctrl) · line 1256

Quoted 10 lines
drivers/nvme/host/core.c +9 -0
 
+/*
+ * Convert integer values from ioctl structures to user pointers, silently
+ * ignoring the upper bits in the compat case to match behaviour of 32-bit
+ * kernels.
+ */
+static void __user *nvme_to_user_ptr(uintptr_t ptrval)
+{
+       if (in_compat_syscall())
+               ptrval = (compat_uptr_t)ptrval;

arm64 make modules failed while building with an extra kernel config.

CONFIG_ARM64_64K_PAGES=y

 # make -sk KBUILD_BUILD_USER=TuxBuild -C/linux -j16 ARCH=arm64
CROSS_COMPILE=aarch64-linux-gnu- HOSTCC=gcc CC="sccache
aarch64-linux-gnu-gcc" O=build modules
70 #
71 ../drivers/nvme/host/core.c: In function ‘nvme_to_user_ptr’:
72 ../drivers/nvme/host/core.c:1256:13: error: ‘compat_uptr_t’
undeclared (first use in this function); did you mean ‘compat_time_t’?
73  1256 | ptrval = (compat_uptr_t)ptrval;
74  | ^~~~~~~~~~~~~
75  | compat_time_t
76 ../drivers/nvme/host/core.c:1256:13: note: each undeclared
identifier is reported only once for each function it appears in
77 ../drivers/nvme/host/core.c:1256:27: error: expected ‘;’ before ‘ptrval’
78  1256 | ptrval = (compat_uptr_t)ptrval;
79  | ^~~~~~
80  | ;
81 make[4]: *** [../scripts/Makefile.build:266:
drivers/nvme/host/core.o] Error 1

full build log,
https://gitlab.com/Linaro/lkft/kernel-runs/-/jobs/528723851

Kernel config:
https://builds.tuxbuild.com/DeL6EepmdRw6OaOGmg7F_g/kernel.config

#39 lore raw

Re: [PATCH AUTOSEL 5.4 38/78] nvme: fix compat address handling in several ioctls

Quoted 4 lines · Naresh Kamboju
Quoted 3 lines · Sasha Levin

From: Nick Bowler <[email protected]>

[ Upstream commit c95b708d5fa65b4e51f088ee077d127fd5a57b70 ]

[...]

Quoted 6 lines

drivers/nvme/host/core.c static void nvme_enable_aen(struct nvme_ctrl *ctrl) · line 1256

Quoted 4 lines
drivers/nvme/host/core.c +4 -0
+static void __user *nvme_to_user_ptr(uintptr_t ptrval)
+{
+       if (in_compat_syscall())
+               ptrval = (compat_uptr_t)ptrval;

arm64 make modules failed while building with an extra kernel config.

[...]

Quoted 2 lines

72 ../drivers/nvme/host/core.c:1256:13: error: ‘compat_uptr_t’ undeclared (first use in this function); did you mean ‘compat_time_t’?

Probably commit 556d687a4ccd5 ("compat: ARM64: always include asm-generic/compat.h") is required to be backported to 5.4 as a prerequisite for including this fix in the 5.4 stable series.

Cheers,
Nick

author maintainer
#40 lore raw

PATCH 39/78 pwm: renesas-tpu: Fix late Runtime PM enablement

In reply to Sasha Levin

From: Geert Uytterhoeven <[email protected]>

[ Upstream commit d5a3c7a4536e1329a758e14340efd0e65252bd3d ]

Runtime PM should be enabled before calling pwmchip_add(), as PWM users can appear immediately after the PWM chip has been added. Likewise, Runtime PM should always be disabled after the removal of the PWM chip, even if the latter failed.

Fixes 99b82abb0a35b073 ("pwm: Add Renesas TPU PWM driver")
---
 drivers/pwm/pwm-renesas-tpu.c | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)
drivers/pwm/pwm-renesas-tpu.c +4 -5
index 4a855a21b782d..8032acc84161a 100644
@@ -415,16 +415,17 @@ static int tpu_probe(struct platform_device *pdev)
 	tpu->chip.base = -1;
 	tpu->chip.npwm = TPU_CHANNEL_MAX;
 
+	pm_runtime_enable(&pdev->dev);
+
 	ret = pwmchip_add(&tpu->chip);
 	if (ret < 0) {
 		dev_err(&pdev->dev, "failed to register PWM chip\n");
+		pm_runtime_disable(&pdev->dev);
 		return ret;
 	}
 
 	dev_info(&pdev->dev, "TPU PWM %d registered\n", tpu->pdev->id);
 
-	pm_runtime_enable(&pdev->dev);
-
 	return 0;
 }
 
@@ -434,12 +435,10 @@ static int tpu_remove(struct platform_device *pdev)
 	int ret;
 
 	ret = pwmchip_remove(&tpu->chip);
-	if (ret)
-		return ret;
 
 	pm_runtime_disable(&pdev->dev);
 
-	return 0;
+	return ret;
 }
 
 #ifdef CONFIG_OF
Signature · 1 line
2.20.1
author maintainer
#41 lore raw

PATCH 40/78 pwm: bcm2835: Dynamically allocate base

In reply to Sasha Levin

From: Florian Fainelli <[email protected]>

[ Upstream commit 2c25b07e5ec119cab609e41407a1fb3fa61442f5 ]

The newer 2711 and 7211 chips have two PWM controllers and failure to dynamically allocate the PWM base would prevent the second PWM controller instance being probed for succeeding with an -EEXIST error from alloc_pwms().

---
 drivers/pwm/pwm-bcm2835.c | 1 +
 1 file changed, 1 insertion(+)
drivers/pwm/pwm-bcm2835.c +1 -0
index 91e24f01b54ed..d78f86f8e4621 100644
@@ -166,6 +166,7 @@ static int bcm2835_pwm_probe(struct platform_device *pdev)
 
 	pc->chip.dev = &pdev->dev;
 	pc->chip.ops = &bcm2835_pwm_ops;
+	pc->chip.base = -1;
 	pc->chip.npwm = 2;
 	pc->chip.of_xlate = of_pwm_xlate_with_flags;
 	pc->chip.of_pwm_n_cells = 3;
Signature · 1 line
2.20.1
author maintainer
#42 lore raw

PATCH 41/78 drm/vc4: Fix HDMI mode validation

In reply to Sasha Levin

From: Nicolas Saenz Julienne <[email protected]>

[ Upstream commit b1e7396a1d0e6af6806337fdaaa44098d6b3343c ]

Current mode validation impedes setting up some video modes which should be supported otherwise. Namely 1920x1200@60Hz.

Fix this by lowering the minimum HDMI state machine clock to pixel clock ratio allowed.

---
 drivers/gpu/drm/vc4/vc4_hdmi.c | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)
drivers/gpu/drm/vc4/vc4_hdmi.c +16 -4
index 0853b980bcb31..d5f5ba4105241 100644
@@ -681,11 +681,23 @@ static enum drm_mode_status
 vc4_hdmi_encoder_mode_valid(struct drm_encoder *crtc,
 			    const struct drm_display_mode *mode)
 {
-	/* HSM clock must be 108% of the pixel clock.  Additionally,
-	 * the AXI clock needs to be at least 25% of pixel clock, but
-	 * HSM ends up being the limiting factor.
+	/*
+	 * As stated in RPi's vc4 firmware "HDMI state machine (HSM) clock must
+	 * be faster than pixel clock, infinitesimally faster, tested in
+	 * simulation. Otherwise, exact value is unimportant for HDMI
+	 * operation." This conflicts with bcm2835's vc4 documentation, which
+	 * states HSM's clock has to be at least 108% of the pixel clock.
+	 *
+	 * Real life tests reveal that vc4's firmware statement holds up, and
+	 * users are able to use pixel clocks closer to HSM's, namely for
+	 * 1920x1200@60Hz. So it was decided to have leave a 1% margin between
+	 * both clocks. Which, for RPi0-3 implies a maximum pixel clock of
+	 * 162MHz.
+	 *
+	 * Additionally, the AXI clock needs to be at least 25% of
+	 * pixel clock, but HSM ends up being the limiting factor.
 	 */
-	if (mode->clock > HSM_CLOCK_FREQ / (1000 * 108 / 100))
+	if (mode->clock > HSM_CLOCK_FREQ / (1000 * 101 / 100))
 		return MODE_CLOCK_HIGH;
 
 	return MODE_OK;
Signature · 6 lines
2.20.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author maintainer
#43 lore raw

PATCH 42/78 iommu/virtio: Fix freeing of incomplete domains

In reply to Sasha Levin

From: Jean-Philippe Brucker <[email protected]>

[ Upstream commit 7062af3ed2ba451029e3733d9f677c68f5ea9e77 ]

Calling viommu_domain_free() on a domain that hasn't been finalised (not attached to any device, for example) can currently cause an Oops, because we attempt to call ida_free() on ID 0, which may either be unallocated or used by another domain.

Only initialise the vdomain->viommu pointer, which denotes a finalised domain, at the end of a successful viommu_domain_finalise().

---
 drivers/iommu/virtio-iommu.c | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)
drivers/iommu/virtio-iommu.c +9 -7
index 3ea9d76829995..6c340a4f4fd28 100644
@@ -614,18 +614,20 @@ static int viommu_domain_finalise(struct viommu_dev *viommu,
 	int ret;
 	struct viommu_domain *vdomain = to_viommu_domain(domain);
 
-	vdomain->viommu		= viommu;
-	vdomain->map_flags	= viommu->map_flags;
+	ret = ida_alloc_range(&viommu->domain_ids, viommu->first_domain,
+			      viommu->last_domain, GFP_KERNEL);
+	if (ret < 0)
+		return ret;
+
+	vdomain->id		= (unsigned int)ret;
 
 	domain->pgsize_bitmap	= viommu->pgsize_bitmap;
 	domain->geometry	= viommu->geometry;
 
-	ret = ida_alloc_range(&viommu->domain_ids, viommu->first_domain,
-			      viommu->last_domain, GFP_KERNEL);
-	if (ret >= 0)
-		vdomain->id = (unsigned int)ret;
+	vdomain->map_flags	= viommu->map_flags;
+	vdomain->viommu		= viommu;
 
-	return ret > 0 ? 0 : ret;
+	return 0;
 }
 
 static void viommu_domain_free(struct iommu_domain *domain)
Signature · 1 line
2.20.1
author maintainer
#44 lore raw

PATCH 43/78 iommu/vt-d: Silence RCU-list debugging warning in dmar_find_atsr()

In reply to Sasha Levin

From: Qian Cai <[email protected]>

[ Upstream commit c6f4ebdeba4cff590594df931ff1ee610c426431 ]

dmar_find_atsr() calls list_for_each_entry_rcu() outside of an RCU read side critical section but with dmar_global_lock held. Silence this false positive.

drivers/iommu/intel-iommu.c:4504 RCU-list traversed in non-reader section!! 1 lock held by swapper/0/1:
#0: ffffffff9755bee8 (dmar_global_lock){+.+.}, at: intel_iommu_init+0x1a6/0xe19

 Call Trace:
  dump_stack+0xa4/0xfe
  lockdep_rcu_suspicious+0xeb/0xf5
  dmar_find_atsr+0x1ab/0x1c0
  dmar_parse_one_atsr+0x64/0x220
  dmar_walk_remapping_entries+0x130/0x380
  dmar_table_init+0x166/0x243
  intel_iommu_init+0x1ab/0xe19
  pci_iommu_init+0x1a/0x44
  do_one_initcall+0xae/0x4d0
  kernel_init_freeable+0x412/0x4c5
  kernel_init+0x19/0x193
---
 drivers/iommu/intel-iommu.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/iommu/intel-iommu.c +2 -1
index 0d922eeae3579..773ac2b0d6068 100644
@@ -4335,7 +4335,8 @@ static struct dmar_atsr_unit *dmar_find_atsr(struct acpi_dmar_atsr *atsr)
 	struct dmar_atsr_unit *atsru;
 	struct acpi_dmar_atsr *tmp;
 
-	list_for_each_entry_rcu(atsru, &dmar_atsr_units, list) {
+	list_for_each_entry_rcu(atsru, &dmar_atsr_units, list,
+				dmar_rcu_check()) {
 		tmp = (struct acpi_dmar_atsr *)atsru->hdr;
 		if (atsr->segment != tmp->segment)
 			continue;
Signature · 6 lines
2.20.1

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
author maintainer
#45 lore raw

PATCH 44/78 platform/chrome: cros_ec_rpmsg: Fix race with host event

In reply to Sasha Levin

From: Pi-Hsun Shih <[email protected]>

[ Upstream commit f775ac78fcfc6bdc96bdda07029d11f2a5e84869 ]

Host event can be sent by remoteproc by any time, and cros_ec_rpmsg_callback would be called after cros_ec_rpmsg_create_ept. But the cros_ec_device is initialized after that, which cause host event handler to use cros_ec_device that are not initialized properly yet.

Fix this by don't schedule host event handler before cros_ec_register returns. Instead, remember that we have a pending host event, and schedule host event handler after cros_ec_register.

Fixes 71cddb7097e2 ("platform/chrome: cros_ec_rpmsg: Fix race with host command when probe failed.")
---
 drivers/platform/chrome/cros_ec_rpmsg.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)
drivers/platform/chrome/cros_ec_rpmsg.c +15 -1
index 0c3738c3244d8..aa2059b33ba56 100644
@@ -42,6 +42,8 @@ struct cros_ec_rpmsg {
 	struct completion xfer_ack;
 	struct work_struct host_event_work;
 	struct rpmsg_endpoint *ept;
+	bool has_pending_host_event;
+	bool probe_done;
 };
 
 /**
@@ -186,7 +188,14 @@ static int cros_ec_rpmsg_callback(struct rpmsg_device *rpdev, void *data,
 		memcpy(ec_dev->din, resp->data, len);
 		complete(&ec_rpmsg->xfer_ack);
 	} else if (resp->type == HOST_EVENT_MARK) {
-		schedule_work(&ec_rpmsg->host_event_work);
+		/*
+		 * If the host event is sent before cros_ec_register is
+		 * finished, queue the host event.
+		 */
+		if (ec_rpmsg->probe_done)
+			schedule_work(&ec_rpmsg->host_event_work);
+		else
+			ec_rpmsg->has_pending_host_event = true;
 	} else {
 		dev_warn(ec_dev->dev, "rpmsg received invalid type = %d",
 			 resp->type);
@@ -249,6 +258,11 @@ static int cros_ec_rpmsg_probe(struct rpmsg_device *rpdev)
 		return ret;
 	}
 
+	ec_rpmsg->probe_done = true;
+
+	if (ec_rpmsg->has_pending_host_event)
+		schedule_work(&ec_rpmsg->host_event_work);
+
 	return 0;
 }
 
Signature · 1 line
2.20.1
author maintainer
#46 lore raw

PATCH 45/78 libnvdimm: Out of bounds read in __nd_ioctl()

In reply to Sasha Levin

From: Dan Carpenter <[email protected]>

[ Upstream commit f84afbdd3a9e5e10633695677b95422572f920dc ]

The "cmd" comes from the user and it can be up to 255. It it's more than the number of bits in long, it results out of bounds read when we check test_bit(cmd, &cmd_mask). The highest valid value for "cmd" is ND_CMD_CALL (10) so I added a compare against that.

Fixes 62232e45f4a2 ("libnvdimm: control (ioctl) messages for nvdimm_bus and nvdimm devices")
---
 drivers/nvdimm/bus.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/nvdimm/bus.c +4 -2
index d47412dcdf38d..5e5c6aafc070b 100644
@@ -1010,8 +1010,10 @@ static int __nd_ioctl(struct nvdimm_bus *nvdimm_bus, struct nvdimm *nvdimm,
 			return -EFAULT;
 	}
 
-	if (!desc || (desc->out_num + desc->in_num == 0) ||
-			!test_bit(cmd, &cmd_mask))
+	if (!desc ||
+	    (desc->out_num + desc->in_num == 0) ||
+	    cmd > ND_CMD_CALL ||
+	    !test_bit(cmd, &cmd_mask))
 		return -ENOTTY;
 
 	/* fail write commands (when read-only) */
Signature · 4 lines
2.20.1
_______________________________________________
Linux-nvdimm mailing list -- [email protected]
To unsubscribe send an email to [email protected]
author maintainer
#47 lore raw

PATCH 46/78 ocfs2: no need try to truncate file beyond i_size

In reply to Sasha Levin

From: Changwei Ge <[email protected]>

[ Upstream commit 783fda856e1034dee90a873f7654c418212d12d7 ]

Linux fallocate(2) with FALLOC_FL_PUNCH_HOLE mode set, its offset can exceed the inode size. Ocfs2 now doesn't allow that offset beyond inode size. This restriction is not necessary and violates fallocate(2) semantics.

If fallocate(2) offset is beyond inode size, just return success and do nothing further.

Otherwise, ocfs2 will crash the kernel.

kernel BUG at fs/ocfs2//alloc.c:7264!

   ocfs2_truncate_inline+0x20f/0x360 [ocfs2]
   ocfs2_remove_inode_range+0x23c/0xcb0 [ocfs2]
   __ocfs2_change_file_space+0x4a5/0x650 [ocfs2]
   ocfs2_fallocate+0x83/0xa0 [ocfs2]
   vfs_fallocate+0x148/0x230
   SyS_fallocate+0x48/0x80
   do_syscall_64+0x79/0x170
Cc Mark Fasheh <[email protected]>
Cc Joel Becker <[email protected]>
Cc Junxiao Bi <[email protected]>
Cc Changwei Ge <[email protected]>
Cc Gang He <[email protected]>
Cc Jun Piao <[email protected]>
Link https://urldefense.com/v3/__http://lkml.kernel.org/r/20200407082754.17565-1-chge at linux.alibaba.com__;!!GqivPVa7Brio!Jo-Ixc0yvMLBBrs3YWIiOQ4YMLqHBfYmk_FdA3GaZXPSfAEXRhtnXVD3MtSIJ7mLfXgk0Q$
---
 fs/ocfs2/alloc.c | 4 ++++
 1 file changed, 4 insertions(+)
fs/ocfs2/alloc.c +4 -0
index f9baefc76cf9b..0a6fe7d5aba74 100644
@@ -7403,6 +7403,10 @@ int ocfs2_truncate_inline(struct inode *inode, struct buffer_head *di_bh,
 	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
 	struct ocfs2_inline_data *idata = &di->id2.i_data;
 
+	/* No need to punch hole beyond i_size. */
+	if (start >= i_size_read(inode))
+		return 0;
+
 	if (end > i_size_read(inode))
 		end = i_size_read(inode);
 
Signature · 1 line
2.20.1
author maintainer
#48 lore raw

PATCH 47/78 hfsplus: fix crash and filesystem corruption when deleting files

In reply to Sasha Levin

From: Simon Gander <[email protected]>

[ Upstream commit 25efb2ffdf991177e740b2f63e92b4ec7d310a92 ]

When removing files containing extended attributes, the hfsplus driver may remove the wrong entries from the attributes b-tree, causing major filesystem damage and in some cases even kernel crashes.

To remove a file, all its extended attributes have to be removed as well. The driver does this by looking up all keys in the attributes b-tree with the cnid of the file. Each of these entries then gets deleted using the key used for searching, which doesn't contain the attribute's name when it should. Since the key doesn't contain the name, the deletion routine will not find the correct entry and instead remove the one in front of it. If parent nodes have to be modified, these become corrupt as well. This causes invalid links and unsorted entries that not even macOS's fsck_hfs is able to fix.

To fix this, modify the search key before an entry is deleted from the attributes b-tree by copying the found entry's key into the search key, therefore ensuring that the correct entry gets removed from the tree.

---
 fs/hfsplus/attributes.c | 4 ++++
 1 file changed, 4 insertions(+)
fs/hfsplus/attributes.c +4 -0
index e6d554476db41..eeebe80c6be4a 100644
@@ -292,6 +292,10 @@ static int __hfsplus_delete_attr(struct inode *inode, u32 cnid,
 		return -ENOENT;
 	}
 
+	/* Avoid btree corruption */
+	hfs_bnode_read(fd->bnode, fd->search_key,
+			fd->keyoffset, fd->keylength);
+
 	err = hfs_brec_remove(fd);
 	if (err)
 		return err;
Signature · 1 line
2.20.1
author maintainer
#49 lore raw

PATCH 48/78 ALSA: hda: Add driver blacklist

In reply to Sasha Levin

From: Takashi Iwai <[email protected]>

[ Upstream commit 3c6fd1f07ed03a04debbb9a9d782205f1ef5e2ab ]

The recent AMD platform exposes an HD-audio bus but without any actual codecs, which is internally tied with a USB-audio device, supposedly. It results in "no codecs" error of HD-audio bus driver, and it's nothing but a waste of resources.

This patch introduces a static blacklist table for skipping such a known bogus PCI SSID entry. As of writing this patch, the known SSIDs are:
* 1043:874f - ASUS ROG Zenith II / Strix
* 1462:cb59 - MSI TRX40 Creator
* 1462:cb60 - MSI TRX40

---
 sound/pci/hda/hda_intel.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
sound/pci/hda/hda_intel.c +16 -0
index 85beb172d8108..1db9d0579c72a 100644
@@ -2024,6 +2024,17 @@ static void pcm_mmap_prepare(struct snd_pcm_substream *substream,
 #endif
 }
 
+/* Blacklist for skipping the whole probe:
+ * some HD-audio PCI entries are exposed without any codecs, and such devices
+ * should be ignored from the beginning.
+ */
+static const struct snd_pci_quirk driver_blacklist[] = {
+	SND_PCI_QUIRK(0x1043, 0x874f, "ASUS ROG Zenith II / Strix", 0),
+	SND_PCI_QUIRK(0x1462, 0xcb59, "MSI TRX40 Creator", 0),
+	SND_PCI_QUIRK(0x1462, 0xcb60, "MSI TRX40", 0),
+	{}
+};
+
 static const struct hda_controller_ops pci_hda_ops = {
 	.disable_msi_reset_irq = disable_msi_reset_irq,
 	.pcm_mmap_prepare = pcm_mmap_prepare,
@@ -2059,6 +2070,11 @@ static int azx_probe(struct pci_dev *pci,
 	bool schedule_probe;
 	int err;
 
+	if (snd_pci_quirk_lookup(pci, driver_blacklist)) {
+		dev_info(&pci->dev, "Skipping the blacklisted device\n");
+		return -ENODEV;
+	}
+
 	if (dev >= SNDRV_CARDS)
 		return -ENODEV;
 	if (!enable[dev]) {
Signature · 1 line
2.20.1
author maintainer
#50 lore raw

PATCH 49/78 ALSA: hda/realtek - Add quirk for MSI GL63

In reply to Sasha Levin

From: Takashi Iwai <[email protected]>

[ Upstream commit 1d3aa4a5516d2e4933fe3cca11d3349ef63bc547 ]

MSI GL63 laptop requires the similar quirk like other MSI models, ALC1220_FIXUP_CLEVO_P950. The board BIOS doesn't provide a PCI SSID for the device, hence we need to take the codec SSID (1462:1275) instead.

---
 sound/pci/hda/patch_realtek.c | 1 +
 1 file changed, 1 insertion(+)
sound/pci/hda/patch_realtek.c +1 -0
index 128db2e6bc64e..190c1aa6a4e62 100644
@@ -2449,6 +2449,7 @@ static const struct snd_pci_quirk alc882_fixup_tbl[] = {
 	SND_PCI_QUIRK(0x1458, 0xa0b8, "Gigabyte AZ370-Gaming", ALC1220_FIXUP_GB_DUAL_CODECS),
 	SND_PCI_QUIRK(0x1458, 0xa0cd, "Gigabyte X570 Aorus Master", ALC1220_FIXUP_CLEVO_P950),
 	SND_PCI_QUIRK(0x1462, 0x1228, "MSI-GP63", ALC1220_FIXUP_CLEVO_P950),
+	SND_PCI_QUIRK(0x1462, 0x1275, "MSI-GL63", ALC1220_FIXUP_CLEVO_P950),
 	SND_PCI_QUIRK(0x1462, 0x1276, "MSI-GL73", ALC1220_FIXUP_CLEVO_P950),
 	SND_PCI_QUIRK(0x1462, 0x1293, "MSI-GP65", ALC1220_FIXUP_CLEVO_P950),
 	SND_PCI_QUIRK(0x1462, 0x7350, "MSI-7350", ALC889_FIXUP_CD),
Signature · 1 line
2.20.1
author maintainer
#51 lore raw

PATCH 50/78 perf/core: Disable page faults when getting phys address

In reply to Sasha Levin

From: Jiri Olsa <[email protected]>

[ Upstream commit d3296fb372bf7497b0e5d0478c4e7a677ec6f6e9 ]

We hit following warning when running tests on kernel compiled with CONFIG_DEBUG_ATOMIC_SLEEP=y:

 WARNING: CPU: 19 PID: 4472 at mm/gup.c:2381 __get_user_pages_fast+0x1a4/0x200
 CPU: 19 PID: 4472 Comm: dummy Not tainted 5.6.0-rc6+ #3
 RIP: 0010:__get_user_pages_fast+0x1a4/0x200
 ...
 Call Trace:
  perf_prepare_sample+0xff1/0x1d90
  perf_event_output_forward+0xe8/0x210
  __perf_event_overflow+0x11a/0x310
  __intel_pmu_pebs_event+0x657/0x850
  intel_pmu_drain_pebs_nhm+0x7de/0x11d0
  handle_pmi_common+0x1b2/0x650
  intel_pmu_handle_irq+0x17b/0x370
  perf_event_nmi_handler+0x40/0x60
  nmi_handle+0x192/0x590
  default_do_nmi+0x6d/0x150
  do_nmi+0x2f9/0x3c0
  nmi+0x8e/0xd7

While __get_user_pages_fast() is IRQ-safe, it calls access_ok(), which warns on:

  WARN_ON_ONCE(!in_task() && !pagefault_disabled())

Peter suggested disabling page faults around __get_user_pages_fast(), which gets rid of the warning in access_ok() call.

---
 kernel/events/core.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)
kernel/events/core.c +6 -3
index 15b123bdcaf53..72d0cfd73cf11 100644
@@ -6537,9 +6537,12 @@ static u64 perf_virt_to_phys(u64 virt)
 		 * Try IRQ-safe __get_user_pages_fast first.
 		 * If failed, leave phys_addr as 0.
 		 */
-		if ((current->mm != NULL) &&
-		    (__get_user_pages_fast(virt, 1, 0, &p) == 1))
-			phys_addr = page_to_phys(p) + virt % PAGE_SIZE;
+		if (current->mm != NULL) {
+			pagefault_disable();
+			if (__get_user_pages_fast(virt, 1, 0, &p) == 1)
+				phys_addr = page_to_phys(p) + virt % PAGE_SIZE;
+			pagefault_enable();
+		}
 
 		if (p)
 			put_page(p);
Signature · 1 line
2.20.1
author maintainer
#52 lore raw

PATCH 51/78 libata: Return correct status in sata_pmp_eh_recover_pm() when ATA_DFLAG_DETACH is set

In reply to Sasha Levin

From: Kai-Heng Feng <[email protected]>

[ Upstream commit 8305f72f952cff21ce8109dc1ea4b321c8efc5af ]

During system resume from suspend, this can be observed on ASM1062 PMP controller:

ata10.01: SATA link down (SStatus 0 SControl 330)
ata10.02: hard resetting link
ata10.02: SATA link down (SStatus 0 SControl 330)
ata10.00: configured for UDMA/133

Kernel panic - not syncing: stack-protector: Kernel
 in: sata_pmp_eh_recover+0xa2b/0xa40

CPU: 2 PID: 230 Comm: scsi_eh_9 Tainted: P OE
#49-Ubuntu
Hardware name: System manufacturer System Product
 1001 12/10/2017
Call Trace:
dump_stack+0x63/0x8b
panic+0xe4/0x244
? sata_pmp_eh_recover+0xa2b/0xa40
__stack_chk_fail+0x19/0x20
sata_pmp_eh_recover+0xa2b/0xa40
? ahci_do_softreset+0x260/0x260 [libahci]
? ahci_do_hardreset+0x140/0x140 [libahci]
? ata_phys_link_offline+0x60/0x60
? ahci_stop_engine+0xc0/0xc0 [libahci]
sata_pmp_error_handler+0x22/0x30
ahci_error_handler+0x45/0x80 [libahci]
ata_scsi_port_error_handler+0x29b/0x770
? ata_scsi_cmd_error_handler+0x101/0x140
ata_scsi_error+0x95/0xd0
? scsi_try_target_reset+0x90/0x90
scsi_error_handler+0xd0/0x5b0
kthread+0x121/0x140
? scsi_eh_get_sense+0x200/0x200
? kthread_create_worker_on_cpu+0x70/0x70
ret_from_fork+0x22/0x40

Kernel Offset: 0xcc00000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff)

Since sata_pmp_eh_recover_pmp() doens't set rc when ATA_DFLAG_DETACH is set, sata_pmp_eh_recover() continues to run. During retry it triggers the stack protector.

Set correct rc in sata_pmp_eh_recover_pmp() to let sata_pmp_eh_recover() jump to pmp_fail directly.

---
 drivers/ata/libata-pmp.c | 1 +
 1 file changed, 1 insertion(+)
drivers/ata/libata-pmp.c +1 -0
index 3ff14071617cd..79f2aeeb482ab 100644
@@ -763,6 +763,7 @@ static int sata_pmp_eh_recover_pmp(struct ata_port *ap,
 
 	if (dev->flags & ATA_DFLAG_DETACH) {
 		detach = 1;
+		rc = -ENODEV;
 		goto fail;
 	}
 
Signature · 1 line
2.20.1
author maintainer
#53 lore raw

PATCH 52/78 drm/amd/display: Calculate scaling ratios on every medium/full update

In reply to Sasha Levin

From: Nicholas Kazlauskas <[email protected]>

[ Upstream commit 3bae20137cae6c03f58f96c0bc9f3d46f0bc17d4 ]

[Why]
If a plane isn't being actively enabled or disabled then DC won't always recalculate scaling rects and ratios for the primary plane.

This results in only a partial or corrupted rect being displayed on the screen instead of scaling to fit the screen.

[How]
Add back the logic to recalculate the scaling rects into dc_commit_updates_for_stream since this is the expected place to do it in DC.

This was previously removed a few years ago to fix an underscan issue but underscan is still functional now with this change - and it should be, since this is only updating to the latest plane state getting passed in.

---
 drivers/gpu/drm/amd/display/dc/core/dc.c | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)
drivers/gpu/drm/amd/display/dc/core/dc.c +12 -1
index 89bd0ba3db1df..71c574d1e8be2 100644
@@ -2154,7 +2154,7 @@ void dc_commit_updates_for_stream(struct dc *dc,
 	enum surface_update_type update_type;
 	struct dc_state *context;
 	struct dc_context *dc_ctx = dc->ctx;
-	int i;
+	int i, j;
 
 	stream_status = dc_stream_get_status(stream);
 	context = dc->current_state;
@@ -2192,6 +2192,17 @@ void dc_commit_updates_for_stream(struct dc *dc,
 
 		copy_surface_update_to_plane(surface, &srf_updates[i]);
 
+		if (update_type >= UPDATE_TYPE_MED) {
+			for (j = 0; j < dc->res_pool->pipe_count; j++) {
+				struct pipe_ctx *pipe_ctx =
+					&context->res_ctx.pipe_ctx[j];
+
+				if (pipe_ctx->plane_state != surface)
+					continue;
+
+				resource_build_scaling_params(pipe_ctx);
+			}
+		}
 	}
 
 	copy_stream_update_to_stream(dc, context, stream, stream_update);
Signature · 6 lines
2.20.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
author maintainer
#54 lore raw

PATCH 53/78 ALSA: ice1724: Fix invalid access for enumerated ctl items

In reply to Sasha Levin

From: Takashi Iwai <[email protected]>

[ Upstream commit c47914c00be346bc5b48c48de7b0da5c2d1a296c ]

The access to Analog Capture Source control value implemented in prodigy_hifi.c is wrong, as caught by the recently introduced sanity check; it should be accessing value.enumerated.item[] instead of value.integer.value[]. This patch corrects the wrong access pattern.

Fixes 6b8d6e5518e2 ("[ALSA] ICE1724: Added support for Audiotrak Prodigy 7.1 HiFi & HD2, Hercules Fortissimo IV")
---
 sound/pci/ice1712/prodigy_hifi.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
sound/pci/ice1712/prodigy_hifi.c +2 -2
index 9d71e9d5c9a06..3cf41c11a4050 100644
@@ -536,7 +536,7 @@ static int wm_adc_mux_enum_get(struct snd_kcontrol *kcontrol,
 	struct snd_ice1712 *ice = snd_kcontrol_chip(kcontrol);
 
 	mutex_lock(&ice->gpio_mutex);
-	ucontrol->value.integer.value[0] = wm_get(ice, WM_ADC_MUX) & 0x1f;
+	ucontrol->value.enumerated.item[0] = wm_get(ice, WM_ADC_MUX) & 0x1f;
 	mutex_unlock(&ice->gpio_mutex);
 	return 0;
 }
@@ -550,7 +550,7 @@ static int wm_adc_mux_enum_put(struct snd_kcontrol *kcontrol,
 
 	mutex_lock(&ice->gpio_mutex);
 	oval = wm_get(ice, WM_ADC_MUX);
-	nval = (oval & 0xe0) | ucontrol->value.integer.value[0];
+	nval = (oval & 0xe0) | ucontrol->value.enumerated.item[0];
 	if (nval != oval) {
 		wm_put(ice, WM_ADC_MUX, nval);
 		change = 1;
Signature · 1 line
2.20.1
author maintainer
#55 lore raw

PATCH 54/78 ALSA: hda: Fix potential access overflow in beep helper

In reply to Sasha Levin

From: Takashi Iwai <[email protected]>

[ Upstream commit 0ad3f0b384d58f3bd1f4fb87d0af5b8f6866f41a ]

The beep control helper function blindly stores the values in two stereo channels no matter whether the actual control is mono or stereo. This is practically harmless, but it annoys the recently introduced sanity check, resulting in an error when the checker is enabled.

This patch corrects the behavior to store only on the defined array member.

---
 sound/pci/hda/hda_beep.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
sound/pci/hda/hda_beep.c +5 -1
index b7d9160ed8688..c6e1e03a5e4da 100644
@@ -290,8 +290,12 @@ int snd_hda_mixer_amp_switch_get_beep(struct snd_kcontrol *kcontrol,
 {
 	struct hda_codec *codec = snd_kcontrol_chip(kcontrol);
 	struct hda_beep *beep = codec->beep;
+	int chs = get_amp_channels(kcontrol);
+
 	if (beep && (!beep->enabled || !ctl_has_mute(kcontrol))) {
-		ucontrol->value.integer.value[0] =
+		if (chs & 1)
+			ucontrol->value.integer.value[0] = beep->enabled;
+		if (chs & 2)
 			ucontrol->value.integer.value[1] = beep->enabled;
 		return 0;
 	}
Signature · 1 line
2.20.1
author maintainer
#56 lore raw

PATCH 55/78 KVM: s390: vsie: Fix delivery of addressing exceptions

In reply to Sasha Levin

From: David Hildenbrand <[email protected]>

[ Upstream commit 4d4cee96fb7a3cc53702a9be8299bf525be4ee98 ]

Whenever we get an -EFAULT, we failed to read in guest 2 physical address space. Such addressing exceptions are reported via a program intercept to the nested hypervisor.

We faked the intercept, we have to return to guest 2. Instead, right now we would be returning -EFAULT from the intercept handler, eventually crashing the VM.
the correct thing to do is to return 1 as rc == 1 is the internal representation of "we have to go back into g2".

Addressing exceptions can only happen if the g2->g3 page tables reference invalid g2 addresses (say, either a table or the final page is not accessible - so something that basically never happens in sane environments.

Identified by manual code inspection.

Fixes a3508fbe9dc6 ("KVM: s390: vsie: initial support for nested virtualization")
Cc <[email protected]> # v4.8+
[[email protected]: fix patch description]
---
 arch/s390/kvm/vsie.c | 1 +
 1 file changed, 1 insertion(+)
arch/s390/kvm/vsie.c +1 -0
index 076090f9e666c..4f6c22d72072a 100644
@@ -1202,6 +1202,7 @@ static int vsie_run(struct kvm_vcpu *vcpu, struct vsie_page *vsie_page)
 		scb_s->iprcc = PGM_ADDRESSING;
 		scb_s->pgmilc = 4;
 		scb_s->gpsw.addr = __rewind_psw(scb_s->gpsw, 4);
+		rc = 1;
 	}
 	return rc;
 }
Signature · 1 line
2.20.1
author maintainer
#57 lore raw

PATCH 56/78 ASoC: Intel: bytcr_rt5640: Add quirk for MPMAN MPWIN895CL tablet

In reply to Sasha Levin

From: Hans de Goede <[email protected]>

[ Upstream commit c8b78f24c1247b7bd0882885c672d9dec5800bc6 ]

The MPMAN MPWIN895CL tablet almost fully works with out default settings. The only problem is that it has only 1 speaker so any sounds only playing on the right channel get lost.

Add a quirk for this model using the default settings + MONO_SPEAKER.

---
 sound/soc/intel/boards/bytcr_rt5640.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
sound/soc/intel/boards/bytcr_rt5640.c +11 -0
index 243f683bc02a7..e62e1d7815aa9 100644
@@ -591,6 +591,17 @@ static const struct dmi_system_id byt_rt5640_quirk_table[] = {
 					BYT_RT5640_SSP0_AIF1 |
 					BYT_RT5640_MCLK_EN),
 	},
+	{
+		/* MPMAN MPWIN895CL */
+		.matches = {
+			DMI_EXACT_MATCH(DMI_SYS_VENDOR, "MPMAN"),
+			DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "MPWIN8900CL"),
+		},
+		.driver_data = (void *)(BYTCR_INPUT_DEFAULTS |
+					BYT_RT5640_MONO_SPEAKER |
+					BYT_RT5640_SSP0_AIF1 |
+					BYT_RT5640_MCLK_EN),
+	},
 	{	/* MSI S100 tablet */
 		.matches = {
 			DMI_EXACT_MATCH(DMI_SYS_VENDOR, "Micro-Star International Co., Ltd."),
Signature · 1 line
2.20.1
author maintainer
#58 lore raw

PATCH 57/78 ipmi: fix hung processes in __get_guid()

In reply to Sasha Levin

From: Wen Yang <[email protected]>

[ Upstream commit 32830a0534700f86366f371b150b17f0f0d140d7 ]

The wait_event() function is used to detect command completion. When send_guid_cmd() returns an error, smi_send() has not been called to send data. Therefore, wait_event() should not be used on the error path, otherwise it will cause the following warning:

[ 1361.588808] systemd-udevd   D    0  1501   1436 0x00000004
[ 1361.588813]  ffff883f4b1298c0 0000000000000000 ffff883f4b188000 ffff887f7e3d9f40
[ 1361.677952]  ffff887f64bd4280 ffffc90037297a68 ffffffff8173ca3b ffffc90000000010
[ 1361.767077]  00ffc90037297ad0 ffff887f7e3d9f40 0000000000000286 ffff883f4b188000
[ 1361.856199] Call Trace:
[ 1361.885578]  [<ffffffff8173ca3b>] ? __schedule+0x23b/0x780
[ 1361.951406]  [<ffffffff8173cfb6>] schedule+0x36/0x80
[ 1362.010979]  [<ffffffffa071f178>] get_guid+0x118/0x150 [ipmi_msghandler]
[ 1362.091281]  [<ffffffff810d5350>] ? prepare_to_wait_event+0x100/0x100
[ 1362.168533]  [<ffffffffa071f755>] ipmi_register_smi+0x405/0x940 [ipmi_msghandler]
[ 1362.258337]  [<ffffffffa0230ae9>] try_smi_init+0x529/0x950 [ipmi_si]
[ 1362.334521]  [<ffffffffa022f350>] ? std_irq_setup+0xd0/0xd0 [ipmi_si]
[ 1362.411701]  [<ffffffffa0232bd2>] init_ipmi_si+0x492/0x9e0 [ipmi_si]
[ 1362.487917]  [<ffffffffa0232740>] ? ipmi_pci_probe+0x280/0x280 [ipmi_si]
[ 1362.568219]  [<ffffffff810021a0>] do_one_initcall+0x50/0x180
[ 1362.636109]  [<ffffffff812231b2>] ? kmem_cache_alloc_trace+0x142/0x190
[ 1362.714330]  [<ffffffff811b2ae1>] do_init_module+0x5f/0x200
[ 1362.781208]  [<ffffffff81123ca8>] load_module+0x1898/0x1de0
[ 1362.848069]  [<ffffffff811202e0>] ? __symbol_put+0x60/0x60
[ 1362.913886]  [<ffffffff8130696b>] ? security_kernel_post_read_file+0x6b/0x80
[ 1362.998514]  [<ffffffff81124465>] SYSC_finit_module+0xe5/0x120
[ 1363.068463]  [<ffffffff81124465>] ? SYSC_finit_module+0xe5/0x120
[ 1363.140513]  [<ffffffff811244be>] SyS_finit_module+0xe/0x10
[ 1363.207364]  [<ffffffff81003c04>] do_syscall_64+0x74/0x180
Fixes 50c812b2b951 ("[PATCH] ipmi: add full sysfs support")
Cc Corey Minyard <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc [email protected] # 2.6.17-

Message-Id: <[email protected]>

---
 drivers/char/ipmi/ipmi_msghandler.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/char/ipmi/ipmi_msghandler.c +2 -2
index 0b6e7f8d97297..90f5292e2051c 100644
@@ -3207,8 +3207,8 @@ static void __get_guid(struct ipmi_smi *intf)
 	if (rv)
 		/* Send failed, no GUID available. */
 		bmc->dyn_guid_set = 0;
-
-	wait_event(intf->waitq, bmc->dyn_guid_set != 2);
+	else
+		wait_event(intf->waitq, bmc->dyn_guid_set != 2);
 
 	/* dyn_guid_set makes the guid data available. */
 	smp_rmb();
Signature · 1 line
2.20.1
author maintainer
#59 lore raw

PATCH 58/78 ALSA: usb-audio: Add Pioneer DJ DJM-250MK2 quirk

In reply to Sasha Levin

From: František Kučera <[email protected]>

[ Upstream commit 73d8c94084341e2895169a0462dbc18167f01683 ]

Pioneer DJ DJM-250MK2 is a mixer that acts like a USB sound card. The MIDI controller part is standard but the PCM part is "vendor specific". Output is enabled by this quirk: 8 channels, 48 000 Hz, S24_3LE. Input is not working.

---
 sound/usb/quirks-table.h | 42 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 42 insertions(+)
sound/usb/quirks-table.h +42 -0
index d187aa6d50db0..dcaf9eed9a415 100644
@@ -3592,5 +3592,47 @@ AU0828_DEVICE(0x2040, 0x7270, "Hauppauge", "HVR-950Q"),
 		}
 	}
 },
+{
+	/*
+	 * Pioneer DJ DJM-250MK2
+	 * PCM is 8 channels out @ 48 fixed (endpoints 0x01).
+	 * The output from computer to the mixer is usable.
+	 *
+	 * The input (phono or line to computer) is not working.
+	 * It should be at endpoint 0x82 and probably also 8 channels,
+	 * but it seems that it works only with Pioneer proprietary software.
+	 * Even on officially supported OS, the Audacity was unable to record
+	 * and Mixxx to recognize the control vinyls.
+	 */
+	USB_DEVICE_VENDOR_SPEC(0x2b73, 0x0017),
+	.driver_info = (unsigned long) &(const struct snd_usb_audio_quirk) {
+		.ifnum = QUIRK_ANY_INTERFACE,
+		.type = QUIRK_COMPOSITE,
+		.data = (const struct snd_usb_audio_quirk[]) {
+			{
+				.ifnum = 0,
+				.type = QUIRK_AUDIO_FIXED_ENDPOINT,
+				.data = &(const struct audioformat) {
+					.formats = SNDRV_PCM_FMTBIT_S24_3LE,
+					.channels = 8, // outputs
+					.iface = 0,
+					.altsetting = 1,
+					.altset_idx = 1,
+					.endpoint = 0x01,
+					.ep_attr = USB_ENDPOINT_XFER_ISOC|
+						USB_ENDPOINT_SYNC_ASYNC,
+					.rates = SNDRV_PCM_RATE_48000,
+					.rate_min = 48000,
+					.rate_max = 48000,
+					.nr_rates = 1,
+					.rate_table = (unsigned int[]) { 48000 }
+				}
+			},
+			{
+				.ifnum = -1
+			}
+		}
+	}
+},
 
 #undef USB_DEVICE_VENDOR_SPEC
Signature · 1 line
2.20.1
author maintainer
#60 lore raw

PATCH 59/78 ALSA: hda/realtek - Add quirk for Lenovo Carbon X1 8th gen

In reply to Sasha Levin

From: Hans de Goede <[email protected]>

[ Upstream commit ca707b3f00b4f31a6e1eb37e8ae99f15f2bb1fe5 ]

The audio setup on the Lenovo Carbon X1 8th gen is the same as that on the Lenovo Carbon X1 7th gen, as such it needs the same ALC285_FIXUP_THINKPAD_HEADSET_JACK quirk.

This fixes volume control of the speaker not working among other things.

---
 sound/pci/hda/patch_realtek.c | 1 +
 1 file changed, 1 insertion(+)
sound/pci/hda/patch_realtek.c +1 -0
index 190c1aa6a4e62..7c327a37eba10 100644
@@ -7300,6 +7300,7 @@ static const struct snd_pci_quirk alc269_fixup_tbl[] = {
 	SND_PCI_QUIRK(0x17aa, 0x225d, "Thinkpad T480", ALC269_FIXUP_LIMIT_INT_MIC_BOOST),
 	SND_PCI_QUIRK(0x17aa, 0x2292, "Thinkpad X1 Yoga 7th", ALC285_FIXUP_THINKPAD_HEADSET_JACK),
 	SND_PCI_QUIRK(0x17aa, 0x2293, "Thinkpad X1 Carbon 7th", ALC285_FIXUP_THINKPAD_HEADSET_JACK),
+	SND_PCI_QUIRK(0x17aa, 0x22be, "Thinkpad X1 Carbon 8th", ALC285_FIXUP_THINKPAD_HEADSET_JACK),
 	SND_PCI_QUIRK(0x17aa, 0x30bb, "ThinkCentre AIO", ALC233_FIXUP_LENOVO_LINE2_MIC_HOTKEY),
 	SND_PCI_QUIRK(0x17aa, 0x30e2, "ThinkCentre AIO", ALC233_FIXUP_LENOVO_LINE2_MIC_HOTKEY),
 	SND_PCI_QUIRK(0x17aa, 0x310c, "ThinkCentre Station", ALC294_FIXUP_LENOVO_MIC_LOCATION),
Signature · 1 line
2.20.1
author maintainer
#61 lore raw

PATCH 60/78 scsi: mpt3sas: Fix kernel panic observed on soft HBA unplug

In reply to Sasha Levin

From: Sreekanth Reddy <[email protected]>

[ Upstream commit cc41f11a21a51d6869d71e525a7264c748d7c0d7 ]

Generic protection fault type kernel panic is observed when user performs soft (ordered) HBA unplug operation while IOs are running on drives connected to HBA.

When user performs ordered HBA removal operation, the kernel calls PCI device's .remove() call back function where driver is flushing out all the outstanding SCSI IO commands with DID_NO_CONNECT host byte and also unmaps sg buffers allocated for these IO commands.

However, in the ordered HBA removal case (unlike of real HBA hot removal), HBA device is still alive and hence HBA hardware is performing the DMA operations to those buffers on the system memory which are already unmapped while flushing out the outstanding SCSI IO commands and this leads to kernel panic.

Don't flush out the outstanding IOs from .remove() path in case of ordered removal since HBA will be still alive in this case and it can complete the outstanding IOs. Flush out the outstanding IOs only in case of 'physical HBA hot unplug' where there won't be any communication with the HBA.

During shutdown also it is possible that HBA hardware can perform DMA operations on those outstanding IO buffers which are completed with DID_NO_CONNECT by the driver from .shutdown(). So same above fix is applied in shutdown path as well.

It is safe to drop the outstanding commands when HBA is inaccessible such as when permanent PCI failure happens, when HBA is in non-operational state, or when someone does a real HBA hot unplug operation. Since driver knows that HBA is inaccessible during these cases, it is safe to drop the outstanding commands instead of waiting for SCSI error recovery to kick in and clear these outstanding commands.

Fixes c666d3be99c0 ("scsi: mpt3sas: wait for and flush running commands on shutdown/unload")
Cc [email protected] #v4.14.174+
---
 drivers/scsi/mpt3sas/mpt3sas_scsih.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/scsi/mpt3sas/mpt3sas_scsih.c +4 -4
index c8e512ba6d398..aff630fccb07e 100644
@@ -9747,8 +9747,8 @@ static void scsih_remove(struct pci_dev *pdev)
 
 	ioc->remove_host = 1;
 
-	mpt3sas_wait_for_commands_to_complete(ioc);
-	_scsih_flush_running_cmds(ioc);
+	if (!pci_device_is_present(pdev))
+		_scsih_flush_running_cmds(ioc);
 
 	_scsih_fw_event_cleanup_queue(ioc);
 
@@ -9831,8 +9831,8 @@ scsih_shutdown(struct pci_dev *pdev)
 
 	ioc->remove_host = 1;
 
-	mpt3sas_wait_for_commands_to_complete(ioc);
-	_scsih_flush_running_cmds(ioc);
+	if (!pci_device_is_present(pdev))
+		_scsih_flush_running_cmds(ioc);
 
 	_scsih_fw_event_cleanup_queue(ioc);
 
Signature · 1 line
2.20.1
author maintainer
#62 lore raw

PATCH 61/78 xhci: Ensure link state is U3 after setting USB_SS_PORT_LS_U3

In reply to Sasha Levin

From: Kai-Heng Feng <[email protected]>

[ Upstream commit eb002726fac7cefb98ff39ddb89e150a1c24fe85 ]

The xHCI spec doesn't specify the upper bound of U3 transition time. For some devices 20ms is not enough, so we need to make sure the link state is in U3 before further actions.

I've tried to use U3 Entry Capability by setting U3 Entry Enable in config register, however the port change event for U3 transition interrupts the system suspend process.

For now let's use the less ideal method by polling PLS.

[use usleep_range(), and shorten the delay time while polling -Mathias]
---
 drivers/usb/host/xhci-hub.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
drivers/usb/host/xhci-hub.c +10 -1
index af92b2576fe91..712cd44f05ace 100644
@@ -1322,7 +1322,16 @@ int xhci_hub_control(struct usb_hcd *hcd, u16 typeReq, u16 wValue,
 			xhci_set_link_state(xhci, ports[wIndex], link_state);
 
 			spin_unlock_irqrestore(&xhci->lock, flags);
-			msleep(20); /* wait device to enter */
+			if (link_state == USB_SS_PORT_LS_U3) {
+				int retries = 16;
+
+				while (retries--) {
+					usleep_range(4000, 8000);
+					temp = readl(ports[wIndex]->addr);
+					if ((temp & PORT_PLS_MASK) == XDEV_U3)
+						break;
+				}
+			}
 			spin_lock_irqsave(&xhci->lock, flags);
 
 			temp = readl(ports[wIndex]->addr);
Signature · 1 line
2.20.1
author maintainer
#63 lore raw

PATCH 62/78 xhci: Wait until link state trainsits to U0 after setting USB_SS_PORT_LS_U0

In reply to Sasha Levin

From: Kai-Heng Feng <[email protected]>

[ Upstream commit 0200b9f790b0fc9e9a42f685f5ad54b23fe959f4 ]

Like U3 case, xHCI spec doesn't specify the upper bound of U0 transition time. The 20ms is not enough for some devices.

Intead of polling PLS or PLC, we can facilitate the port change event to know that the link transits to U0 is completed.

While at it, also separate U0 and U3 case to make the code cleaner.

[variable rename to u3exit, and skip completion for usb2 ports -Mathias ]
---
 drivers/usb/host/xhci-hub.c  | 44 +++++++++++++++++++++++++-----------
 drivers/usb/host/xhci-mem.c  |  1 +
 drivers/usb/host/xhci-ring.c |  1 +
 drivers/usb/host/xhci.h      |  1 +
 4 files changed, 34 insertions(+), 13 deletions(-)
drivers/usb/host/xhci-hub.c +31 -13
index 712cd44f05ace..02f52d4f74df8 100644
@@ -1306,7 +1306,33 @@ int xhci_hub_control(struct usb_hcd *hcd, u16 typeReq, u16 wValue,
 					 wIndex, link_state);
 				goto error;
 			}
+
+			if (link_state == USB_SS_PORT_LS_U0) {
+				if ((temp & PORT_PLS_MASK) == XDEV_U0)
+					break;
+
+				if (!((temp & PORT_PLS_MASK) == XDEV_U1 ||
+				    (temp & PORT_PLS_MASK) == XDEV_U2 ||
+				    (temp & PORT_PLS_MASK) == XDEV_U3)) {
+					xhci_warn(xhci, "Can only set port %d to U0 from U state\n",
+							wIndex);
+					goto error;
+				}
+				reinit_completion(&bus_state->u3exit_done[wIndex]);
+				xhci_set_link_state(xhci, ports[wIndex],
+						    USB_SS_PORT_LS_U0);
+				spin_unlock_irqrestore(&xhci->lock, flags);
+				if (!wait_for_completion_timeout(&bus_state->u3exit_done[wIndex],
+								 msecs_to_jiffies(100)))
+					xhci_dbg(xhci, "missing U0 port change event for port %d\n",
+						 wIndex);
+				spin_lock_irqsave(&xhci->lock, flags);
+				temp = readl(ports[wIndex]->addr);
+				break;
+			}
+
 			if (link_state == USB_SS_PORT_LS_U3) {
+				int retries = 16;
 				slot_id = xhci_find_slot_id_by_port(hcd, xhci,
 						wIndex + 1);
 				if (slot_id) {
@@ -1317,26 +1343,18 @@ int xhci_hub_control(struct usb_hcd *hcd, u16 typeReq, u16 wValue,
 					xhci_stop_device(xhci, slot_id, 1);
 					spin_lock_irqsave(&xhci->lock, flags);
 				}
-			}
-
-			xhci_set_link_state(xhci, ports[wIndex], link_state);
-
-			spin_unlock_irqrestore(&xhci->lock, flags);
-			if (link_state == USB_SS_PORT_LS_U3) {
-				int retries = 16;
-
+				xhci_set_link_state(xhci, ports[wIndex], USB_SS_PORT_LS_U3);
+				spin_unlock_irqrestore(&xhci->lock, flags);
 				while (retries--) {
 					usleep_range(4000, 8000);
 					temp = readl(ports[wIndex]->addr);
 					if ((temp & PORT_PLS_MASK) == XDEV_U3)
 						break;
 				}
-			}
-			spin_lock_irqsave(&xhci->lock, flags);
-
-			temp = readl(ports[wIndex]->addr);
-			if (link_state == USB_SS_PORT_LS_U3)
+				spin_lock_irqsave(&xhci->lock, flags);
+				temp = readl(ports[wIndex]->addr);
 				bus_state->suspended_ports |= 1 << wIndex;
+			}
 			break;
 		case USB_PORT_FEAT_POWER:
 			/*
drivers/usb/host/xhci-mem.c +1 -0
index 884c601bfa15f..9764122c9cdf2 100644
@@ -2552,6 +2552,7 @@ int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags)
 		xhci->usb3_rhub.bus_state.resume_done[i] = 0;
 		/* Only the USB 2.0 completions will ever be used. */
 		init_completion(&xhci->usb2_rhub.bus_state.rexit_done[i]);
+		init_completion(&xhci->usb3_rhub.bus_state.u3exit_done[i]);
 	}
 
 	if (scratchpad_alloc(xhci, flags))
drivers/usb/host/xhci-ring.c +1 -0
index f7a190fb2353d..a952d99f0456b 100644
@@ -1673,6 +1673,7 @@ static void handle_port_status(struct xhci_hcd *xhci,
 	     (portsc & PORT_PLS_MASK) == XDEV_U1 ||
 	     (portsc & PORT_PLS_MASK) == XDEV_U2)) {
 		xhci_dbg(xhci, "resume SS port %d finished\n", port_id);
+		complete(&bus_state->u3exit_done[hcd_portnum]);
 		/* We've just brought the device into U0/1/2 through either the
 		 * Resume state after a device remote wakeup, or through the
 		 * U3Exit state after a host-initiated resume.  If it's a device
drivers/usb/host/xhci.h +1 -0
index 98b98a0cd2a8b..3ad4f80960d08 100644
@@ -1694,6 +1694,7 @@ struct xhci_bus_state {
 	/* Which ports are waiting on RExit to U0 transition. */
 	unsigned long		rexit_ports;
 	struct completion	rexit_done[USB_MAXCHILDREN];
+	struct completion	u3exit_done[USB_MAXCHILDREN];
 };
 
 
Signature · 1 line
2.20.1
author maintainer
#64 lore raw

PATCH 63/78 xhci: Finetune host initiated USB3 rootport link suspend and resume

In reply to Sasha Levin

From: Mathias Nyman <[email protected]>

[ Upstream commit ceca49382ac20e06ce04c21279c7f2868c4ec1d4 ]

Depending on the current link state the steps to resume the link to U0 varies. The normal case when a port is suspended (U3) we set the link to U0 and wait for a port event when U3exit completed and port moved to U0.

If the port is in U1/U2, then no event is issued, just set link to U0

If port is in Resume or Recovery state then the device has already initiated resume, and this host initiated resume is racing against it. Port event handler for device initiated resume will set link to U0, just wait for the port to reach U0 before returning.

---
 drivers/usb/host/xhci-hub.c | 36 +++++++++++++++++++++++++-----------
 1 file changed, 25 insertions(+), 11 deletions(-)
drivers/usb/host/xhci-hub.c +25 -11
index 02f52d4f74df8..a9c87eb8951e8 100644
@@ -1307,20 +1307,34 @@ int xhci_hub_control(struct usb_hcd *hcd, u16 typeReq, u16 wValue,
 				goto error;
 			}
 
+			/*
+			 * set link to U0, steps depend on current link state.
+			 * U3: set link to U0 and wait for u3exit completion.
+			 * U1/U2:  no PLC complete event, only set link to U0.
+			 * Resume/Recovery: device initiated U0, only wait for
+			 * completion
+			 */
 			if (link_state == USB_SS_PORT_LS_U0) {
-				if ((temp & PORT_PLS_MASK) == XDEV_U0)
-					break;
+				u32 pls = temp & PORT_PLS_MASK;
+				bool wait_u0 = false;
 
-				if (!((temp & PORT_PLS_MASK) == XDEV_U1 ||
-				    (temp & PORT_PLS_MASK) == XDEV_U2 ||
-				    (temp & PORT_PLS_MASK) == XDEV_U3)) {
-					xhci_warn(xhci, "Can only set port %d to U0 from U state\n",
-							wIndex);
-					goto error;
+				/* already in U0 */
+				if (pls == XDEV_U0)
+					break;
+				if (pls == XDEV_U3 ||
+				    pls == XDEV_RESUME ||
+				    pls == XDEV_RECOVERY) {
+					wait_u0 = true;
+					reinit_completion(&bus_state->u3exit_done[wIndex]);
+				}
+				if (pls <= XDEV_U3) /* U1, U2, U3 */
+					xhci_set_link_state(xhci, ports[wIndex],
+							    USB_SS_PORT_LS_U0);
+				if (!wait_u0) {
+					if (pls > XDEV_U3)
+						goto error;
+					break;
 				}
-				reinit_completion(&bus_state->u3exit_done[wIndex]);
-				xhci_set_link_state(xhci, ports[wIndex],
-						    USB_SS_PORT_LS_U0);
 				spin_unlock_irqrestore(&xhci->lock, flags);
 				if (!wait_for_completion_timeout(&bus_state->u3exit_done[wIndex],
 								 msecs_to_jiffies(100)))
Signature · 1 line
2.20.1
author maintainer
#65 lore raw

PATCH 64/78 drm/amd/display: Not doing optimize bandwidth if flip pending.

In reply to Sasha Levin

From: Yongqiang Sun <[email protected]>

[ Upstream commit 9941b8129030c9202aaf39114477a0e58c0d6ffc ]

[Why]
In some scenario like 1366x768 VSR enabled connected with a 4K monitor and playing 4K video in clone mode, underflow will be observed due to decrease dppclk when previouse surface scan isn't finished

[How]
In this use case, surface flip is switching between 4K and 1366x768, 1366x768 needs smaller dppclk, and when decrease the clk and previous surface scan is for 4K and scan isn't done, underflow will happen. Not doing optimize bandwidth in case of flip pending.

---
 drivers/gpu/drm/amd/display/dc/core/dc.c | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)
drivers/gpu/drm/amd/display/dc/core/dc.c +23 -0
index 71c574d1e8be2..2028dc017f7a0 100644
@@ -1182,6 +1182,26 @@ bool dc_commit_state(struct dc *dc, struct dc_state *context)
 	return (result == DC_OK);
 }
 
+static bool is_flip_pending_in_pipes(struct dc *dc, struct dc_state *context)
+{
+	int i;
+	struct pipe_ctx *pipe;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (!pipe->plane_state)
+			continue;
+
+		/* Must set to false to start with, due to OR in update function */
+		pipe->plane_state->status.is_flip_pending = false;
+		dc->hwss.update_pending_status(pipe);
+		if (pipe->plane_state->status.is_flip_pending)
+			return true;
+	}
+	return false;
+}
+
 bool dc_post_update_surfaces_to_stream(struct dc *dc)
 {
 	int i;
@@ -1192,6 +1212,9 @@ bool dc_post_update_surfaces_to_stream(struct dc *dc)
 
 	post_surface_trace(dc);
 
+	if (is_flip_pending_in_pipes(dc, context))
+		return true;
+
 	for (i = 0; i < dc->res_pool->pipe_count; i++)
 		if (context->res_ctx.pipe_ctx[i].stream == NULL ||
 		    context->res_ctx.pipe_ctx[i].plane_state == NULL) {
Signature · 6 lines
2.20.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
author maintainer
#66 lore raw

PATCH 65/78 PCI/PM: Add pcie_wait_for_link_delay()

In reply to Sasha Levin

From: Mika Westerberg <[email protected]>

[ Upstream commit 4827d63891b6a839dac49c6ab62e61c4b011c4f2 ]

Add pcie_wait_for_link_delay(). Similar to pcie_wait_for_link() but allows passing custom activation delay in milliseconds.

---
 drivers/pci/pci.c | 21 ++++++++++++++++++---
 1 file changed, 18 insertions(+), 3 deletions(-)
drivers/pci/pci.c +18 -3
index 981ae16f935bc..94d6e120b4734 100644
@@ -4605,14 +4605,17 @@ static int pci_pm_reset(struct pci_dev *dev, int probe)
 
 	return pci_dev_wait(dev, "PM D3->D0", PCIE_RESET_READY_POLL_MS);
 }
+
 /**
- * pcie_wait_for_link - Wait until link is active or inactive
+ * pcie_wait_for_link_delay - Wait until link is active or inactive
  * @pdev: Bridge device
  * @active: waiting for active or inactive?
+ * @delay: Delay to wait after link has become active (in ms)
  *
  * Use this to wait till link becomes active or inactive.
  */
-bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
+static bool pcie_wait_for_link_delay(struct pci_dev *pdev, bool active,
+				     int delay)
 {
 	int timeout = 1000;
 	bool ret;
@@ -4649,13 +4652,25 @@ bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
 		timeout -= 10;
 	}
 	if (active && ret)
-		msleep(100);
+		msleep(delay);
 	else if (ret != active)
 		pci_info(pdev, "Data Link Layer Link Active not %s in 1000 msec\n",
 			active ? "set" : "cleared");
 	return ret == active;
 }
 
+/**
+ * pcie_wait_for_link - Wait until link is active or inactive
+ * @pdev: Bridge device
+ * @active: waiting for active or inactive?
+ *
+ * Use this to wait till link becomes active or inactive.
+ */
+bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
+{
+	return pcie_wait_for_link_delay(pdev, active, 100);
+}
+
 void pci_reset_secondary_bus(struct pci_dev *dev)
 {
 	u16 ctrl;
Signature · 1 line
2.20.1
author maintainer
#67 lore raw

PATCH 66/78 libbpf: Fix readelf output parsing on powerpc with recent binutils

In reply to Sasha Levin

From: Aurelien Jarno <[email protected]>

[ Upstream commit 3464afdf11f9a1e031e7858a05351ceca1792fea ]

On powerpc with recent versions of binutils, readelf outputs an extra field when dumping the symbols of an object file. For example:

    35: 0000000000000838    96 FUNC    LOCAL  DEFAULT [<localentry>: 8]     1 btf_is_struct

The extra "[<localentry>: 8]" prevents the GLOBAL_SYM_COUNT variable to be computed correctly and causes the check_abi target to fail.

Fix that by looking for the symbol name in the last field instead of the 8th one. This way it should also cope with future extra fields.

---
 tools/lib/bpf/Makefile | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
tools/lib/bpf/Makefile +2 -2
index 33e2638ef7f0d..122321d549227 100644
@@ -145,7 +145,7 @@ PC_FILE		:= $(addprefix $(OUTPUT),$(PC_FILE))
 
 GLOBAL_SYM_COUNT = $(shell readelf -s --wide $(BPF_IN_SHARED) | \
 			   cut -d "@" -f1 | sed 's/_v[0-9]_[0-9]_[0-9].*//' | \
-			   awk '/GLOBAL/ && /DEFAULT/ && !/UND/ {print $$8}' | \
+			   awk '/GLOBAL/ && /DEFAULT/ && !/UND/ {print $$NF}' | \
 			   sort -u | wc -l)
 VERSIONED_SYM_COUNT = $(shell readelf -s --wide $(OUTPUT)libbpf.so | \
 			      grep -Eo '[^ ]+@LIBBPF_' | cut -d@ -f1 | sort -u | wc -l)
@@ -217,7 +217,7 @@ check_abi: $(OUTPUT)libbpf.so
 		     "versioned in $(VERSION_SCRIPT)." >&2;		 \
 		readelf -s --wide $(BPF_IN_SHARED) |			 \
 		    cut -d "@" -f1 | sed 's/_v[0-9]_[0-9]_[0-9].*//' |	 \
-		    awk '/GLOBAL/ && /DEFAULT/ && !/UND/ {print $$8}'|   \
+		    awk '/GLOBAL/ && /DEFAULT/ && !/UND/ {print $$NF}'|  \
 		    sort -u > $(OUTPUT)libbpf_global_syms.tmp;		 \
 		readelf -s --wide $(OUTPUT)libbpf.so |			 \
 		    grep -Eo '[^ ]+@LIBBPF_' | cut -d@ -f1 |		 \
Signature · 1 line
2.20.1
author maintainer
#68 lore raw

PATCH 67/78 PCI: pciehp: Prevent deadlock on disconnect

In reply to Sasha Levin

From: Mika Westerberg <[email protected]>

[ Upstream commit 87d0f2a5536fdf5053a6d341880f96135549a644 ]

This addresses deadlocks in these common cases in hierarchies containing two switches:

- All involved ports are runtime suspended and they are unplugged. This can happen easily if the drivers involved automatically enable runtime PM (xHCI for example does that).

- System is suspended (e.g., closing the lid on a laptop) with a dock + something else connected, and the dock is unplugged while suspended.

These cases lead to the following deadlock:

  INFO: task irq/126-pciehp:198 blocked for more than 120 seconds.
  irq/126-pciehp  D    0   198      2 0x80000000
  Call Trace:
   schedule+0x2c/0x80
   schedule_timeout+0x246/0x350
   wait_for_completion+0xb7/0x140
   kthread_stop+0x49/0x110
   free_irq+0x32/0x70
   pcie_shutdown_notification+0x2f/0x50
   pciehp_remove+0x27/0x50
   pcie_port_remove_service+0x36/0x50
   device_release_driver+0x12/0x20
   bus_remove_device+0xec/0x160
   device_del+0x13b/0x350
   device_unregister+0x1a/0x60
   remove_iter+0x1e/0x30
   device_for_each_child+0x56/0x90
   pcie_port_device_remove+0x22/0x40
   pcie_portdrv_remove+0x20/0x60
   pci_device_remove+0x3e/0xc0
   device_release_driver_internal+0x18c/0x250
   device_release_driver+0x12/0x20
   pci_stop_bus_device+0x6f/0x90
   pci_stop_bus_device+0x31/0x90
   pci_stop_and_remove_bus_device+0x12/0x20
   pciehp_unconfigure_device+0x88/0x140
   pciehp_disable_slot+0x6a/0x110
   pciehp_handle_presence_or_link_change+0x263/0x400
   pciehp_ist+0x1c9/0x1d0
   irq_thread_fn+0x24/0x60
   irq_thread+0xeb/0x190
   kthread+0x120/0x140

  INFO: task irq/190-pciehp:2288 blocked for more than 120 seconds.
  irq/190-pciehp  D    0  2288      2 0x80000000
  Call Trace:
   __schedule+0x2a2/0x880
   schedule+0x2c/0x80
   schedule_preempt_disabled+0xe/0x10
   mutex_lock+0x2c/0x30
   pci_lock_rescan_remove+0x15/0x20
   pciehp_unconfigure_device+0x4d/0x140
   pciehp_disable_slot+0x6a/0x110
   pciehp_handle_presence_or_link_change+0x263/0x400
   pciehp_ist+0x1c9/0x1d0
   irq_thread_fn+0x24/0x60
   irq_thread+0xeb/0x190
   kthread+0x120/0x140

What happens here is that the whole hierarchy is runtime resumed and the parent PCIe downstream port, which got the hot-remove event, starts removing devices below it, taking pci_lock_rescan_remove() lock. When the child PCIe port is runtime resumed it calls pciehp_check_presence() which ends up calling pciehp_card_present() and pciehp_check_link_active(). Both of these use pcie_capability_read_word(), which notices that the underlying device is already gone and returns PCIBIOS_DEVICE_NOT_FOUND with the capability value set to 0. When pciehp gets this value it thinks that its child device is also hot-removed and schedules its IRQ thread to handle the event.

The deadlock happens when the child's IRQ thread runs and tries to acquire pci_lock_rescan_remove() which is already taken by the parent and the parent waits for the child's IRQ thread to finish.

Prevent this from happening by checking the return value of pcie_capability_read_word() and if it is PCIBIOS_DEVICE_NOT_FOUND stop performing any hot-removal activities.

---
 drivers/pci/hotplug/pciehp.h      |  6 ++--
 drivers/pci/hotplug/pciehp_core.c | 11 ++++--
 drivers/pci/hotplug/pciehp_ctrl.c |  4 +--
 drivers/pci/hotplug/pciehp_hpc.c  | 59 +++++++++++++++++++++++++------
 4 files changed, 61 insertions(+), 19 deletions(-)
drivers/pci/hotplug/pciehp.h +3 -3
index 882ce82c46990..aa61d4c219d7b 100644
@@ -174,10 +174,10 @@ void pciehp_set_indicators(struct controller *ctrl, int pwr, int attn);
 
 void pciehp_get_latch_status(struct controller *ctrl, u8 *status);
 int pciehp_query_power_fault(struct controller *ctrl);
-bool pciehp_card_present(struct controller *ctrl);
-bool pciehp_card_present_or_link_active(struct controller *ctrl);
+int pciehp_card_present(struct controller *ctrl);
+int pciehp_card_present_or_link_active(struct controller *ctrl);
 int pciehp_check_link_status(struct controller *ctrl);
-bool pciehp_check_link_active(struct controller *ctrl);
+int pciehp_check_link_active(struct controller *ctrl);
 void pciehp_release_ctrl(struct controller *ctrl);
 
 int pciehp_sysfs_enable_slot(struct hotplug_slot *hotplug_slot);
drivers/pci/hotplug/pciehp_core.c +8 -3
index 56daad828c9e0..312cc45c44c78 100644
@@ -139,10 +139,15 @@ static int get_adapter_status(struct hotplug_slot *hotplug_slot, u8 *value)
 {
 	struct controller *ctrl = to_ctrl(hotplug_slot);
 	struct pci_dev *pdev = ctrl->pcie->port;
+	int ret;
 
 	pci_config_pm_runtime_get(pdev);
-	*value = pciehp_card_present_or_link_active(ctrl);
+	ret = pciehp_card_present_or_link_active(ctrl);
 	pci_config_pm_runtime_put(pdev);
+	if (ret < 0)
+		return ret;
+
+	*value = ret;
 	return 0;
 }
 
@@ -158,13 +163,13 @@ static int get_adapter_status(struct hotplug_slot *hotplug_slot, u8 *value)
  */
 static void pciehp_check_presence(struct controller *ctrl)
 {
-	bool occupied;
+	int occupied;
 
 	down_read(&ctrl->reset_lock);
 	mutex_lock(&ctrl->state_lock);
 
 	occupied = pciehp_card_present_or_link_active(ctrl);
-	if ((occupied && (ctrl->state == OFF_STATE ||
+	if ((occupied > 0 && (ctrl->state == OFF_STATE ||
 			  ctrl->state == BLINKINGON_STATE)) ||
 	    (!occupied && (ctrl->state == ON_STATE ||
 			   ctrl->state == BLINKINGOFF_STATE)))
drivers/pci/hotplug/pciehp_ctrl.c +2 -2
index dd8e4a5fb2826..6503d15effbbd 100644
@@ -226,7 +226,7 @@ void pciehp_handle_disable_request(struct controller *ctrl)
 
 void pciehp_handle_presence_or_link_change(struct controller *ctrl, u32 events)
 {
-	bool present, link_active;
+	int present, link_active;
 
 	/*
 	 * If the slot is on and presence or link has changed, turn it off.
@@ -257,7 +257,7 @@ void pciehp_handle_presence_or_link_change(struct controller *ctrl, u32 events)
 	mutex_lock(&ctrl->state_lock);
 	present = pciehp_card_present(ctrl);
 	link_active = pciehp_check_link_active(ctrl);
-	if (!present && !link_active) {
+	if (present <= 0 && link_active <= 0) {
 		mutex_unlock(&ctrl->state_lock);
 		return;
 	}
drivers/pci/hotplug/pciehp_hpc.c +48 -11

88 lines not on this page · raw message

Signature · 1 line
2.20.1
author maintainer
#69 lore raw

PATCH 68/78 ASoC: SOF: trace: fix unconditional free in trace release

In reply to Sasha Levin

From: Kai Vehmanen <[email protected]>

[ Upstream commit e6110114d18d330c05fd6de9f31283fd086a5a3a ]

Check if DMA pages were successfully allocated in initialization before calling free. For many types of memory (like sgbufs) the extra free is harmless, but not all backends track allocation state, so add an explicit check.

---
 sound/soc/sof/trace.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
sound/soc/sof/trace.c +5 -2
index 4c3cff031fd66..fd6f5913782bf 100644
@@ -328,7 +328,10 @@ void snd_sof_free_trace(struct snd_sof_dev *sdev)
 {
 	snd_sof_release_trace(sdev);
 
-	snd_dma_free_pages(&sdev->dmatb);
-	snd_dma_free_pages(&sdev->dmatp);
+	if (sdev->dma_trace_pages) {
+		snd_dma_free_pages(&sdev->dmatb);
+		snd_dma_free_pages(&sdev->dmatp);
+		sdev->dma_trace_pages = 0;
+	}
 }
 EXPORT_SYMBOL(snd_sof_free_trace);
Signature · 1 line
2.20.1
author maintainer
#70 lore raw

PATCH 69/78 powerpc/powernv/ioda: Fix ref count for devices with their own PE

In reply to Sasha Levin

From: Frederic Barrat <[email protected]>

[ Upstream commit 05dd7da76986937fb288b4213b1fa10dbe0d1b33 ]

The pci_dn structure used to store a pointer to the struct pci_dev, so taking a reference on the device was required. However, the pci_dev pointer was later removed from the pci_dn structure, but the reference was kept for the npu device.
See commit 902bdc57451c ("powerpc/powernv/idoa: Remove unnecessary pcidev from pci_dn").

We don't need to take a reference on the device when assigning the PE as the struct pnv_ioda_pe is cleaned up at the same time as the (physical) device is released. Doing so prevents the device from being released, which is a problem for opencapi devices, since we want to be able to remove them through PCI hotplug.

Now the ugly part: nvlink npu devices are not meant to be released. Because of the above, we've always leaked a reference and simply removing it now is dangerous and would likely require more work. There's currently no release device callback for nvlink devices for example. So to be safe, this patch leaks a reference on the npu device, but only for nvlink and not opencapi.

---
 arch/powerpc/platforms/powernv/pci-ioda.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)
arch/powerpc/platforms/powernv/pci-ioda.c +12 -7
index 058223233088e..e9cda7e316a50 100644
@@ -1062,14 +1062,13 @@ static struct pnv_ioda_pe *pnv_ioda_setup_dev_PE(struct pci_dev *dev)
 		return NULL;
 	}
 
-	/* NOTE: We get only one ref to the pci_dev for the pdn, not for the
-	 * pointer in the PE data structure, both should be destroyed at the
-	 * same time. However, this needs to be looked at more closely again
-	 * once we actually start removing things (Hotplug, SR-IOV, ...)
+	/* NOTE: We don't get a reference for the pointer in the PE
+	 * data structure, both the device and PE structures should be
+	 * destroyed at the same time. However, removing nvlink
+	 * devices will need some work.
 	 *
 	 * At some point we want to remove the PDN completely anyways
 	 */
-	pci_dev_get(dev);
 	pdn->pe_number = pe->pe_number;
 	pe->flags = PNV_IODA_PE_DEV;
 	pe->pdev = dev;
@@ -1084,7 +1083,6 @@ static struct pnv_ioda_pe *pnv_ioda_setup_dev_PE(struct pci_dev *dev)
 		pnv_ioda_free_pe(pe);
 		pdn->pe_number = IODA_INVALID_PE;
 		pe->pdev = NULL;
-		pci_dev_put(dev);
 		return NULL;
 	}
 
@@ -1205,6 +1203,14 @@ static struct pnv_ioda_pe *pnv_ioda_setup_npu_PE(struct pci_dev *npu_pdev)
 	struct pci_controller *hose = pci_bus_to_host(npu_pdev->bus);
 	struct pnv_phb *phb = hose->private_data;
 
+	/*
+	 * Intentionally leak a reference on the npu device (for
+	 * nvlink only; this is not an opencapi path) to make sure it
+	 * never goes away, as it's been the case all along and some
+	 * work is needed otherwise.
+	 */
+	pci_dev_get(npu_pdev);
+
 	/*
 	 * Due to a hardware errata PE#0 on the NPU is reserved for
 	 * error handling. This means we only have three PEs remaining
@@ -1228,7 +1234,6 @@ static struct pnv_ioda_pe *pnv_ioda_setup_npu_PE(struct pci_dev *npu_pdev)
 			 */
 			dev_info(&npu_pdev->dev,
 				"Associating to existing PE %x\n", pe_num);
-			pci_dev_get(npu_pdev);
 			npu_pdn = pci_get_pdn(npu_pdev);
 			rid = npu_pdev->bus->number << 8 | npu_pdn->devfn;
 			npu_pdn->pe_number = pe_num;
Signature · 1 line
2.20.1
#71 lore raw

Re: [PATCH AUTOSEL 5.4 69/78] powerpc/powernv/ioda: Fix ref count for devices with their own PE

Le 18/04/2020 à 16:40, Sasha Levin a écrit :

Quoted 3 lines

From: Frederic Barrat <[email protected]>

[ Upstream commit 05dd7da76986937fb288b4213b1fa10dbe0d1b33 ]

This shouldn't be backported to stable.

Fred

Quoted 84 lines

The pci_dn structure used to store a pointer to the struct pci_dev, so taking a reference on the device was required. However, the pci_dev pointer was later removed from the pci_dn structure, but the reference was kept for the npu device.
See commit 902bdc57451c ("powerpc/powernv/idoa: Remove unnecessary pcidev from pci_dn").

We don't need to take a reference on the device when assigning the PE as the struct pnv_ioda_pe is cleaned up at the same time as the (physical) device is released. Doing so prevents the device from being released, which is a problem for opencapi devices, since we want to be able to remove them through PCI hotplug.

Now the ugly part: nvlink npu devices are not meant to be released. Because of the above, we've always leaked a reference and simply removing it now is dangerous and would likely require more work. There's currently no release device callback for nvlink devices for example. So to be safe, this patch leaks a reference on the npu device, but only for nvlink and not opencapi.

---
  arch/powerpc/platforms/powernv/pci-ioda.c | 19 ++++++++++++-------
  1 file changed, 12 insertions(+), 7 deletions(-)
arch/powerpc/platforms/powernv/pci-ioda.c +12 -7
index 058223233088e..e9cda7e316a50 100644
@@ -1062,14 +1062,13 @@ static struct pnv_ioda_pe *pnv_ioda_setup_dev_PE(struct pci_dev *dev)
  		return NULL;
  	}
 
-	/* NOTE: We get only one ref to the pci_dev for the pdn, not for the
-	 * pointer in the PE data structure, both should be destroyed at the
-	 * same time. However, this needs to be looked at more closely again
-	 * once we actually start removing things (Hotplug, SR-IOV, ...)
+	/* NOTE: We don't get a reference for the pointer in the PE
+	 * data structure, both the device and PE structures should be
+	 * destroyed at the same time. However, removing nvlink
+	 * devices will need some work.
  	 *
  	 * At some point we want to remove the PDN completely anyways
  	 */
-	pci_dev_get(dev);
  	pdn->pe_number = pe->pe_number;
  	pe->flags = PNV_IODA_PE_DEV;
  	pe->pdev = dev;
@@ -1084,7 +1083,6 @@ static struct pnv_ioda_pe *pnv_ioda_setup_dev_PE(struct pci_dev *dev)
  		pnv_ioda_free_pe(pe);
  		pdn->pe_number = IODA_INVALID_PE;
  		pe->pdev = NULL;
-		pci_dev_put(dev);
  		return NULL;
  	}
 
@@ -1205,6 +1203,14 @@ static struct pnv_ioda_pe *pnv_ioda_setup_npu_PE(struct pci_dev *npu_pdev)
  	struct pci_controller *hose = pci_bus_to_host(npu_pdev->bus);
  	struct pnv_phb *phb = hose->private_data;
 
+	/*
+	 * Intentionally leak a reference on the npu device (for
+	 * nvlink only; this is not an opencapi path) to make sure it
+	 * never goes away, as it's been the case all along and some
+	 * work is needed otherwise.
+	 */
+	pci_dev_get(npu_pdev);
+
  	/*
  	 * Due to a hardware errata PE#0 on the NPU is reserved for
  	 * error handling. This means we only have three PEs remaining
@@ -1228,7 +1234,6 @@ static struct pnv_ioda_pe *pnv_ioda_setup_npu_PE(struct pci_dev *npu_pdev)
  			 */
  			dev_info(&npu_pdev->dev,
  				"Associating to existing PE %x\n", pe_num);
-			pci_dev_get(npu_pdev);
  			npu_pdn = pci_get_pdn(npu_pdev);
  			rid = npu_pdev->bus->number << 8 | npu_pdn->devfn;
  			npu_pdn->pe_number = pe_num;
author maintainer
#72 lore raw

Re: [PATCH AUTOSEL 5.4 69/78] powerpc/powernv/ioda: Fix ref count for devices with their own PE

Quoted 9 lines · Frederic Barrat

Le 18/04/2020 à 16:40, Sasha Levin a écrit :

Quoted 3 lines

From: Frederic Barrat <[email protected]>

[ Upstream commit 05dd7da76986937fb288b4213b1fa10dbe0d1b33 ]

This shouldn't be backported to stable.

I've dropped this and the other two commits you've pointed out from all branches, thanks!

Signature · 2 lines
Thanks,
Sasha
author maintainer
#73 lore raw

PATCH 70/78 pci/hotplug/pnv-php: Remove erroneous warning

In reply to Sasha Levin

From: Frederic Barrat <[email protected]>

[ Upstream commit 658ab186dd22060408d94f5c5a6d02d809baba44 ]

On powernv, when removing a device through hotplug, the following warning is logged:

     Invalid refcount <.> on <...>

It may be incorrect, the refcount may be set to a higher value than 1 and be valid. of_detach_node() can drop more than one reference. As it doesn't seem trivial to assert the correct value, let's remove the warning.

---
 drivers/pci/hotplug/pnv_php.c | 6 ------
 1 file changed, 6 deletions(-)
drivers/pci/hotplug/pnv_php.c +0 -6
index d7b2b47bc33eb..6037983c6e46b 100644
@@ -151,17 +151,11 @@ static void pnv_php_rmv_pdns(struct device_node *dn)
 static void pnv_php_detach_device_nodes(struct device_node *parent)
 {
 	struct device_node *dn;
-	int refcount;
 
 	for_each_child_of_node(parent, dn) {
 		pnv_php_detach_device_nodes(dn);
 
 		of_node_put(dn);
-		refcount = kref_read(&dn->kobj.kref);
-		if (refcount != 1)
-			pr_warn("Invalid refcount %d on <%pOF>\n",
-				refcount, dn);
-
 		of_detach_node(dn);
 	}
 }
Signature · 1 line
2.20.1
#74 lore raw

Re: [PATCH AUTOSEL 5.4 70/78] pci/hotplug/pnv-php: Remove erroneous warning

Le 18/04/2020 à 16:40, Sasha Levin a écrit :

Quoted 3 lines

From: Frederic Barrat <[email protected]>

[ Upstream commit 658ab186dd22060408d94f5c5a6d02d809baba44 ]

This doesn't need to be backported to stable.

Fred

Quoted 43 lines

On powernv, when removing a device through hotplug, the following warning is logged:

      Invalid refcount <.> on <...>

It may be incorrect, the refcount may be set to a higher value than 1 and be valid. of_detach_node() can drop more than one reference. As it doesn't seem trivial to assert the correct value, let's remove the warning.

---
  drivers/pci/hotplug/pnv_php.c | 6 ------
  1 file changed, 6 deletions(-)
drivers/pci/hotplug/pnv_php.c +0 -6
index d7b2b47bc33eb..6037983c6e46b 100644
@@ -151,17 +151,11 @@ static void pnv_php_rmv_pdns(struct device_node *dn)
  static void pnv_php_detach_device_nodes(struct device_node *parent)
  {
  	struct device_node *dn;
-	int refcount;
 
  	for_each_child_of_node(parent, dn) {
  		pnv_php_detach_device_nodes(dn);
 
  		of_node_put(dn);
-		refcount = kref_read(&dn->kobj.kref);
-		if (refcount != 1)
-			pr_warn("Invalid refcount %d on <%pOF>\n",
-				refcount, dn);
-
  		of_detach_node(dn);
  	}
  }
author maintainer
#75 lore raw

PATCH 71/78 ocxl: Add PCI hotplug dependency to Kconfig

In reply to Sasha Levin

From: Frederic Barrat <[email protected]>

[ Upstream commit 49ce94b8677c7d7a15c4d7cbbb9ff1cd8387827b ]

The PCI hotplug framework is used to update the devices when a new image is written to the FPGA.

---
 drivers/misc/ocxl/Kconfig | 1 +
 1 file changed, 1 insertion(+)
drivers/misc/ocxl/Kconfig +1 -0
index 1916fa65f2f2a..2d2266c1439ef 100644
@@ -11,6 +11,7 @@ config OCXL
 	tristate "OpenCAPI coherent accelerator support"
 	depends on PPC_POWERNV && PCI && EEH
 	select OCXL_BASE
+	select HOTPLUG_PCI_POWERNV
 	default m
 	help
 	  Select this option to enable the ocxl driver for Open
Signature · 1 line
2.20.1
#76 lore raw

Re: [PATCH AUTOSEL 5.4 71/78] ocxl: Add PCI hotplug dependency to Kconfig

Le 18/04/2020 à 16:40, Sasha Levin a écrit :

Quoted 3 lines

From: Frederic Barrat <[email protected]>

[ Upstream commit 49ce94b8677c7d7a15c4d7cbbb9ff1cd8387827b ]

This shouldn't be backported to stable.

Fred

Quoted 26 lines

The PCI hotplug framework is used to update the devices when a new image is written to the FPGA.

---
  drivers/misc/ocxl/Kconfig | 1 +
  1 file changed, 1 insertion(+)
drivers/misc/ocxl/Kconfig +1 -0
index 1916fa65f2f2a..2d2266c1439ef 100644
@@ -11,6 +11,7 @@ config OCXL
  	tristate "OpenCAPI coherent accelerator support"
  	depends on PPC_POWERNV && PCI && EEH
  	select OCXL_BASE
+	select HOTPLUG_PCI_POWERNV
  	default m
  	help
  	  Select this option to enable the ocxl driver for Open
author maintainer
#77 lore raw

PATCH 72/78 tracing/selftests: Turn off timeout setting

In reply to Sasha Levin

From: "Steven Rostedt (VMware)" <[email protected]>

[ Upstream commit b43e78f65b1d35fd3e13c7b23f9b64ea83c9ad3a ]

As the ftrace selftests can run for a long period of time, disable the timeout that the general selftests have. If a selftest hangs, then it probably means the machine will hang too.

---
 tools/testing/selftests/ftrace/settings | 1 +
 1 file changed, 1 insertion(+)
 create mode 100644 tools/testing/selftests/ftrace/settings
tools/testing/selftests/ftrace/settings new file +1 -0
new file mode 100644
index 0000000000000..e7b9417537fbc
@@ -0,0 +1 @@
+timeout=0
Signature · 1 line
2.20.1
author maintainer
#78 lore raw

PATCH 73/78 virtio-blk: improve virtqueue error to BLK_STS

In reply to Sasha Levin

From: Halil Pasic <[email protected]>

[ Upstream commit 3d973b2e9a625996ee997c7303cd793b9d197c65 ]

Let's change the mapping between virtqueue_add errors to BLK_STS statuses, so that -ENOSPC, which indicates virtqueue full is still mapped to BLK_STS_DEV_RESOURCE, but -ENOMEM which indicates non-device specific resource outage is mapped to BLK_STS_RESOURCE.

---
 drivers/block/virtio_blk.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
drivers/block/virtio_blk.c +7 -2
index c2ed3e9128e3a..a55383b139df9 100644
@@ -345,9 +345,14 @@ static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
 		if (err == -ENOSPC)
 			blk_mq_stop_hw_queue(hctx);
 		spin_unlock_irqrestore(&vblk->vqs[qid].lock, flags);
-		if (err == -ENOMEM || err == -ENOSPC)
+		switch (err) {
+		case -ENOSPC:
 			return BLK_STS_DEV_RESOURCE;
-		return BLK_STS_IOERR;
+		case -ENOMEM:
+			return BLK_STS_RESOURCE;
+		default:
+			return BLK_STS_IOERR;
+		}
 	}
 
 	if (bd->last && virtqueue_kick_prepare(vblk->vqs[qid].vq))
Signature · 1 line
2.20.1
author maintainer
#79 lore raw

PATCH 74/78 scsi: smartpqi: fix controller lockup observed during force reboot

In reply to Sasha Levin

From: Kevin Barnett <[email protected]>

[ Upstream commit 0530736e40a0695b1ee2762e2684d00549699da4 ]

---
 drivers/scsi/smartpqi/smartpqi.h      |   9 +-
 drivers/scsi/smartpqi/smartpqi_init.c | 126 ++++++++++++++++++++++----
 2 files changed, 115 insertions(+), 20 deletions(-)
drivers/scsi/smartpqi/smartpqi.h +8 -1
index 79d2af36f6552..2aa81b22f2695 100644
@@ -1130,8 +1130,9 @@ struct pqi_ctrl_info {
 	struct mutex	ofa_mutex; /* serialize ofa */
 	bool		controller_online;
 	bool		block_requests;
-	bool		in_shutdown;
+	bool		block_device_reset;
 	bool		in_ofa;
+	bool		in_shutdown;
 	u8		inbound_spanning_supported : 1;
 	u8		outbound_spanning_supported : 1;
 	u8		pqi_mode_enabled : 1;
@@ -1173,6 +1174,7 @@ struct pqi_ctrl_info {
 	struct          pqi_ofa_memory *pqi_ofa_mem_virt_addr;
 	dma_addr_t      pqi_ofa_mem_dma_handle;
 	void            **pqi_ofa_chunk_virt_addr;
+	atomic_t	sync_cmds_outstanding;
 };
 
 enum pqi_ctrl_mode {
@@ -1423,6 +1425,11 @@ static inline bool pqi_ctrl_blocked(struct pqi_ctrl_info *ctrl_info)
 	return ctrl_info->block_requests;
 }
 
+static inline bool pqi_device_reset_blocked(struct pqi_ctrl_info *ctrl_info)
+{
+	return ctrl_info->block_device_reset;
+}
+
 void pqi_sas_smp_handler(struct bsg_job *job, struct Scsi_Host *shost,
 	struct sas_rphy *rphy);
 
drivers/scsi/smartpqi/smartpqi_init.c +107 -19

229 lines not on this page · raw message

Signature · 1 line
2.20.1
author maintainer
#80 lore raw

PATCH 75/78 scsi: smartpqi: fix call trace in device discovery

In reply to Sasha Levin

From: Murthy Bhat <[email protected]>

[ Upstream commit b969261134c1b990b96ea98fe5e0fcf8ec937c04 ]

Use sas_phy_delete rather than sas_phy_free which, according to comments, should not be called for PHYs that have been set up successfully.

---
 drivers/scsi/smartpqi/smartpqi_sas_transport.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/scsi/smartpqi/smartpqi_sas_transport.c +1 -1
index 6776dfc1d317c..b7e28b9f8589f 100644
@@ -45,9 +45,9 @@ static void pqi_free_sas_phy(struct pqi_sas_phy *pqi_sas_phy)
 	struct sas_phy *phy = pqi_sas_phy->phy;
 
 	sas_port_delete_phy(pqi_sas_phy->parent_port->port, phy);
-	sas_phy_free(phy);
 	if (pqi_sas_phy->added_to_port)
 		list_del(&pqi_sas_phy->phy_list_entry);
+	sas_phy_delete(phy);
 	kfree(pqi_sas_phy);
 }
 
Signature · 1 line
2.20.1
author maintainer
#81 lore raw

PATCH 76/78 scsi: smartpqi: fix problem with unique ID for physical device

In reply to Sasha Levin

From: Kevin Barnett <[email protected]>

[ Upstream commit 5b083b305b49f65269b888885455b8c0cf1a52e4 ]

Obtain the unique IDs from the RLL and RPL instead of VPD page 83h.

---
 drivers/scsi/smartpqi/smartpqi.h      |  1 -
 drivers/scsi/smartpqi/smartpqi_init.c | 99 ++++-----------------------
 2 files changed, 12 insertions(+), 88 deletions(-)
drivers/scsi/smartpqi/smartpqi.h +0 -1
index 2aa81b22f2695..7a3a942b40df0 100644
@@ -907,7 +907,6 @@ struct pqi_scsi_dev {
 	u8	scsi3addr[8];
 	__be64	wwid;
 	u8	volume_id[16];
-	u8	unique_id[16];
 	u8	is_physical_device : 1;
 	u8	is_external_raid_device : 1;
 	u8	is_expander_smp_device : 1;
drivers/scsi/smartpqi/smartpqi_init.c +12 -87

128 lines not on this page · raw message

Signature · 1 line
2.20.1
author maintainer
#82 lore raw

PATCH 77/78 PCI/ASPM: Allow re-enabling Clock PM

In reply to Sasha Levin

From: Heiner Kallweit <[email protected]>

[ Upstream commit 35efea32b26f9aacc99bf07e0d2cdfba2028b099 ]

Previously Clock PM could not be re-enabled after being disabled by pci_disable_link_state() because clkpm_capable was reset. Change this by adding a clkpm_disable field similar to aspm_disable.

---
 drivers/pci/pcie/aspm.c | 18 +++++++++++-------
 1 file changed, 11 insertions(+), 7 deletions(-)
drivers/pci/pcie/aspm.c +11 -7
index 652ef23bba35a..ccf3a69bd97ff 100644
@@ -64,6 +64,7 @@ struct pcie_link_state {
 	u32 clkpm_capable:1;		/* Clock PM capable? */
 	u32 clkpm_enabled:1;		/* Current Clock PM state */
 	u32 clkpm_default:1;		/* Default Clock PM state by BIOS */
+	u32 clkpm_disable:1;		/* Clock PM disabled */
 
 	/* Exit latencies */
 	struct aspm_latency latency_up;	/* Upstream direction exit latency */
@@ -161,8 +162,11 @@ static void pcie_set_clkpm_nocheck(struct pcie_link_state *link, int enable)
 
 static void pcie_set_clkpm(struct pcie_link_state *link, int enable)
 {
-	/* Don't enable Clock PM if the link is not Clock PM capable */
-	if (!link->clkpm_capable)
+	/*
+	 * Don't enable Clock PM if the link is not Clock PM capable
+	 * or Clock PM is disabled
+	 */
+	if (!link->clkpm_capable || link->clkpm_disable)
 		enable = 0;
 	/* Need nothing if the specified equals to current state */
 	if (link->clkpm_enabled == enable)
@@ -192,7 +196,8 @@ static void pcie_clkpm_cap_init(struct pcie_link_state *link, int blacklist)
 	}
 	link->clkpm_enabled = enabled;
 	link->clkpm_default = enabled;
-	link->clkpm_capable = (blacklist) ? 0 : capable;
+	link->clkpm_capable = capable;
+	link->clkpm_disable = blacklist ? 1 : 0;
 }
 
 static bool pcie_retrain_link(struct pcie_link_state *link)
@@ -1097,10 +1102,9 @@ static int __pci_disable_link_state(struct pci_dev *pdev, int state, bool sem)
 		link->aspm_disable |= ASPM_STATE_L1;
 	pcie_config_aspm_link(link, policy_to_aspm_state(link));
 
-	if (state & PCIE_LINK_STATE_CLKPM) {
-		link->clkpm_capable = 0;
-		pcie_set_clkpm(link, 0);
-	}
+	if (state & PCIE_LINK_STATE_CLKPM)
+		link->clkpm_disable = 1;
+	pcie_set_clkpm(link, policy_to_clkpm_state(link));
 	mutex_unlock(&aspm_lock);
 	if (sem)
 		up_read(&pci_bus_sem);
Signature · 1 line
2.20.1
author maintainer
#83 lore raw

PATCH 78/78 PCI/PM: Add missing link delays required by the PCIe spec

In reply to Sasha Levin

From: Mika Westerberg <[email protected]>

[ Upstream commit ad9001f2f41198784b0423646450ba2cb24793a3 ]

Currently Linux does not follow PCIe spec regarding the required delays after reset. A concrete example is a Thunderbolt add-in-card that consists of a PCIe switch and two PCIe endpoints:

  +-1b.0-[01-6b]----00.0-[02-6b]--+-00.0-[03]----00.0 TBT controller
                                  +-01.0-[04-36]-- DS hotplug port
                                  +-02.0-[37]----00.0 xHCI controller
                                  \-04.0-[38-6b]-- DS hotplug port

The root port (1b.0) and the PCIe switch downstream ports are all PCIe Gen3 so they support 8GT/s link speeds.

We wait for the PCIe hierarchy to enter D3cold (runtime):

pcieport 0000:00:1b.0: power state changed by ACPI to D3cold

When it wakes up from D3cold, according to the PCIe 5.0 section 5.8 the PCIe switch is put to reset and its power is re-applied. This means that we must follow the rules in PCIe 5.0 section 6.6.1.

For the PCIe Gen3 ports we are dealing with here, the following applies:

With a Downstream Port that supports Link speeds greater than 5.0 GT/s, software must wait a minimum of 100 ms after Link training completes before sending a Configuration Request to the device immediately below that Port. Software can determine when Link training completes by polling the Data Link Layer Link Active bit or by setting up an associated interrupt (see Section 6.7.3.3).

Translating this into the above topology we would need to do this (DLLLA stands for Data Link Layer Link Active):

0000:00:1b.0: wait for 100 ms after DLLLA is set before access to 0000:01:00.0 0000:02:00.0: wait for 100 ms after DLLLA is set before access to 0000:03:00.0 0000:02:02.0: wait for 100 ms after DLLLA is set before access to 0000:37:00.0

I've instrumented the kernel with some additional logging so we can see the actual delays performed:

pcieport 0000:00:1b.0: power state changed by ACPI to D0 pcieport 0000:00:1b.0: waiting for D3cold delay of 100 ms pcieport 0000:00:1b.0: waiting for D3hot delay of 10 ms pcieport 0000:02:01.0: waiting for D3hot delay of 10 ms pcieport 0000:02:04.0: waiting for D3hot delay of 10 ms

For the switch upstream port (01:00.0 reachable through 00:1b.0 root port) we wait for 100 ms but not taking into account the DLLLA requirement. We then wait 10 ms for D3hot -> D0 transition of the root port and the two downstream hotplug ports. This means that we deviate from what the spec requires.

Performing the same check for system sleep (s2idle) transitions it turns out to be even worse. None of the mandatory delays are performed. If this would be S3 instead of s2idle then according to PCI FW spec 3.2 section 4.6.8. there is a specific _DSM that allows the OS to skip the delays but this platform does not provide the _DSM and does not go to S3 anyway so no firmware is involved that could already handle these delays.

On this particular platform these delays are not actually needed because there is an additional delay as part of the ACPI power resource that is used to turn on power to the hierarchy but since that additional delay is not required by any of standards (PCIe, ACPI) it is not present in the Intel Ice Lake, for example where missing the mandatory delays causes pciehp to start tearing down the stack too early (links are not yet trained). Below is an example how it looks like when this happens:

pcieport 0000:83:04.0: pciehp: Slot(4): Card not present
pcieport 0000:87:04.0: PME# disabled
pcieport 0000:83:04.0: pciehp: pciehp_unconfigure_device: domain:bus:dev = 0000:86:00 pcieport 0000:86:00.0: Refused to change power state, currently in D3
pcieport 0000:86:00.0: restoring config space at offset 0x3c (was 0xffffffff, writing 0x201ff) pcieport 0000:86:00.0: restoring config space at offset 0x38 (was 0xffffffff, writing 0x0)
...

There is also one reported case (see the bugzilla link below) where the missing delay causes xHCI on a Titan Ridge controller fail to runtime resume when USB-C dock is plugged. This does not involve pciehp but instead the PCI core fails to runtime resume the xHCI device:

pcieport 0000:04:02.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020) pcieport 0000:04:02.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100406) xhci_hcd 0000:39:00.0: Refused to change power state, currently in D3 xhci_hcd 0000:39:00.0: restoring config space at offset 0x3c (was 0xffffffff, writing 0x1ff) xhci_hcd 0000:39:00.0: restoring config space at offset 0x38 (was 0xffffffff, writing 0x0)
...

Add a new function pci_bridge_wait_for_secondary_bus() that is called on PCI core resume and runtime resume paths accordingly if the bridge entered D3cold (and thus went through reset).

This is second attempt to add the missing delays. The previous solution in c2bf1fc212f7 ("PCI: Add missing link delays required by the PCIe spec") was reverted because of two issues it caused:

1. One system become unresponsive after S3 resume due to PME service spinning in pcie_pme_work_fn(). The root port in question reports that the xHCI sent PME but the xHCI device itself does not have PME status set. The PME status bit is never cleared in the root port resulting the indefinite loop in pcie_pme_work_fn().

2. Slows down resume if the root/downstream port does not support Data Link Layer Active Reporting because pcie_wait_for_link_delay() waits 1100 ms in that case.

This version should avoid the above issues because we restrict the delay to happen only if the port went into D3cold.

---
 drivers/pci/pci-driver.c |  11 +++-
 drivers/pci/pci.c        | 121 ++++++++++++++++++++++++++++++++++++++-
 drivers/pci/pci.h        |   1 +
 3 files changed, 130 insertions(+), 3 deletions(-)
drivers/pci/pci-driver.c +10 -1
index 0c3086793e4ec..5ea612a15550e 100644
@@ -919,6 +919,8 @@ static int pci_pm_resume_noirq(struct device *dev)
 	struct pci_dev *pci_dev = to_pci_dev(dev);
 	struct device_driver *drv = dev->driver;
 	int error = 0;
+	pci_power_t prev_state = pci_dev->current_state;
+	bool skip_bus_pm = pci_dev->skip_bus_pm;
 
 	if (dev_pm_may_skip_resume(dev))
 		return 0;
@@ -937,12 +939,15 @@ static int pci_pm_resume_noirq(struct device *dev)
 	 * configuration here and attempting to put them into D0 again is
 	 * pointless, so avoid doing that.
 	 */
-	if (!(pci_dev->skip_bus_pm && pm_suspend_no_platform()))
+	if (!(skip_bus_pm && pm_suspend_no_platform()))
 		pci_pm_default_resume_early(pci_dev);
 
 	pci_fixup_device(pci_fixup_resume_early, pci_dev);
 	pcie_pme_root_status_cleanup(pci_dev);
 
+	if (!skip_bus_pm && prev_state == PCI_D3cold)
+		pci_bridge_wait_for_secondary_bus(pci_dev);
+
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_resume_early(dev);
 
@@ -1333,6 +1338,7 @@ static int pci_pm_runtime_resume(struct device *dev)
 	int rc = 0;
 	struct pci_dev *pci_dev = to_pci_dev(dev);
 	const struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
+	pci_power_t prev_state = pci_dev->current_state;
 
 	/*
 	 * Restoring config space is necessary even if the device is not bound
@@ -1348,6 +1354,9 @@ static int pci_pm_runtime_resume(struct device *dev)
 	pci_enable_wake(pci_dev, PCI_D0, false);
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
+	if (prev_state == PCI_D3cold)
+		pci_bridge_wait_for_secondary_bus(pci_dev);
+
 	if (pm && pm->runtime_resume)
 		rc = pm->runtime_resume(dev);
 
drivers/pci/pci.c +119 -2

133 lines not on this page · raw message

drivers/pci/pci.h +1 -0
index 273d60cb0762d..a5adc2e2c351d 100644
@@ -107,6 +107,7 @@ void pci_allocate_cap_save_buffers(struct pci_dev *dev);
 void pci_free_cap_save_buffers(struct pci_dev *dev);
 bool pci_bridge_d3_possible(struct pci_dev *dev);
 void pci_bridge_d3_update(struct pci_dev *dev);
+void pci_bridge_wait_for_secondary_bus(struct pci_dev *dev);
 
 static inline void pci_wakeup_event(struct pci_dev *dev)
 {
Signature · 1 line
2.20.1