Skip to contentSkip to conversation

netfilter: nfnetlink: avoid deadlock due to synchronous request_module

PATCH In review Sasha Levin 2 participants 31 messages Aug 2, 2019

Subsystems (13)
lore
Share & tools
lore

Sasha Levin

author maintainer
#0 lore raw

PATCH 1/30 netfilter: nfnetlink: avoid deadlock due to synchronous request_module

From: Florian Westphal <[email protected]>

[ Upstream commit 1b0890cd60829bd51455dc5ad689ed58c4408227 ]

Thomas and Juliana report a deadlock when running:

(rmmod nf_conntrack_netlink/xfrm_user)

conntrack -e NEW -E &
modprobe -v xfrm_user

They provided following analysis:

conntrack -e NEW -E

    netlink_bind()
        netlink_lock_table() -> increases "nl_table_users"
            nfnetlink_bind()
            # does not unlock the table as it's locked by netlink_bind()
                __request_module()
                    call_usermodehelper_exec()

This triggers "modprobe nf_conntrack_netlink" from kernel, netlink_bind() won't return until modprobe process is done.

"modprobe xfrm_user":

    xfrm_user_init()
        register_pernet_subsys()
            -> grab pernet_ops_rwsem
                ..
                netlink_table_grab()
                    calls schedule() as "nl_table_users" is non-zero

so modprobe is blocked because netlink_bind() increased nl_table_users while also holding pernet_ops_rwsem.

"modprobe nf_conntrack_netlink" runs and inits nf_conntrack_netlink:

    ctnetlink_init()
        register_pernet_subsys()
            -> blocks on "pernet_ops_rwsem" thanks to xfrm_user module

both modprobe processes wait on one another -- neither can make progress.

Switch netlink_bind() to "nowait" modprobe -- this releases the netlink table lock, which then allows both modprobe instances to complete.

---
 net/netfilter/nfnetlink.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/netfilter/nfnetlink.c +1 -1
index 733d3e4a30d85..2cee032af46d2 100644
@@ -530,7 +530,7 @@ static int nfnetlink_bind(struct net *net, int group)
 	ss = nfnetlink_get_subsys(type << 8);
 	rcu_read_unlock();
 	if (!ss)
-		request_module("nfnetlink-subsys-%d", type);
+		request_module_nowait("nfnetlink-subsys-%d", type);
 	return 0;
 }
 #endif
Signature · 1 line
2.20.1
author maintainer
#1 lore raw

PATCH 2/30 vfio-ccw: Set pa_nr to 0 if memory allocation fails for pa_iova_pfn

From: Farhan Ali <[email protected]>

[ Upstream commit c1ab69268d124ebdbb3864580808188ccd3ea355 ]

So we don't call try to call vfio_unpin_pages() incorrectly.

Fixes 0a19e61e6d4c ("vfio: ccw: introduce channel program interfaces")

Message-Id: <33a89467ad6369196ae6edf820cbcb1e2d8d050c.1562854091.git.alifm@linux.ibm.com>

---
 drivers/s390/cio/vfio_ccw_cp.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/s390/cio/vfio_ccw_cp.c +3 -1
index 1419eaea03d84..5a9e457caef33 100644
@@ -119,8 +119,10 @@ static int pfn_array_alloc_pin(struct pfn_array *pa, struct device *mdev,
 				  sizeof(*pa->pa_iova_pfn) +
 				  sizeof(*pa->pa_pfn),
 				  GFP_KERNEL);
-	if (unlikely(!pa->pa_iova_pfn))
+	if (unlikely(!pa->pa_iova_pfn)) {
+		pa->pa_nr = 0;
 		return -ENOMEM;
+	}
 	pa->pa_pfn = pa->pa_iova_pfn + pa->pa_nr;
 
 	ret = pfn_array_pin(pa, mdev);
Signature · 1 line
2.20.1
author maintainer
#2 lore raw

PATCH 3/30 netfilter: Fix rpfilter dropping vrf packets by mistake

In reply to Sasha Levin

From: Miaohe Lin <[email protected]>

[ Upstream commit b575b24b8eee37f10484e951b62ce2a31c579775 ]

When firewalld is enabled with ipv4/ipv6 rpfilter, vrf ipv4/ipv6 packets will be dropped. Vrf device will pass through netfilter hook twice. One with enslaved device and another one with l3 master device. So in device may dismatch witch out device because out device is always enslaved device.So failed with the check of the rpfilter and drop the packets by mistake.

---
 net/ipv4/netfilter/ipt_rpfilter.c  | 1 +
 net/ipv6/netfilter/ip6t_rpfilter.c | 8 ++++++--
 2 files changed, 7 insertions(+), 2 deletions(-)
net/ipv4/netfilter/ipt_rpfilter.c +1 -0
index 37fb9552e8589..341d1bd637af2 100644
@@ -96,6 +96,7 @@ static bool rpfilter_mt(const struct sk_buff *skb, struct xt_action_param *par)
 	flow.flowi4_mark = info->flags & XT_RPFILTER_VALID_MARK ? skb->mark : 0;
 	flow.flowi4_tos = RT_TOS(iph->tos);
 	flow.flowi4_scope = RT_SCOPE_UNIVERSE;
+	flow.flowi4_oif = l3mdev_master_ifindex_rcu(xt_in(par));
 
 	return rpfilter_lookup_reverse(xt_net(par), &flow, xt_in(par), info->flags) ^ invert;
 }
net/ipv6/netfilter/ip6t_rpfilter.c +6 -2
index 40eb16bd97860..d535768bea0fd 100644
@@ -58,7 +58,9 @@ static bool rpfilter_lookup_reverse6(struct net *net, const struct sk_buff *skb,
 	if (rpfilter_addr_linklocal(&iph->saddr)) {
 		lookup_flags |= RT6_LOOKUP_F_IFACE;
 		fl6.flowi6_oif = dev->ifindex;
-	} else if ((flags & XT_RPFILTER_LOOSE) == 0)
+	/* Set flowi6_oif for vrf devices to lookup route in l3mdev domain. */
+	} else if (netif_is_l3_master(dev) || netif_is_l3_slave(dev) ||
+		  (flags & XT_RPFILTER_LOOSE) == 0)
 		fl6.flowi6_oif = dev->ifindex;
 
 	rt = (void *) ip6_route_lookup(net, &fl6, lookup_flags);
@@ -73,7 +75,9 @@ static bool rpfilter_lookup_reverse6(struct net *net, const struct sk_buff *skb,
 		goto out;
 	}
 
-	if (rt->rt6i_idev->dev == dev || (flags & XT_RPFILTER_LOOSE))
+	if (rt->rt6i_idev->dev == dev ||
+	    l3mdev_master_ifindex_rcu(rt->rt6i_idev->dev) == dev->ifindex ||
+	    (flags & XT_RPFILTER_LOOSE))
 		ret = true;
  out:
 	ip6_rt_put(rt);
Signature · 1 line
2.20.1
author maintainer
#3 lore raw

PATCH 4/30 netfilter: nft_hash: fix symhash with modulus one

In reply to Sasha Levin

From: Laura Garcia Liebana <[email protected]>

[ Upstream commit 28b1d6ef53e3303b90ca8924bb78f31fa527cafb ]

The rule below doesn't work as the kernel raises -ERANGE.

nft add rule netdev nftlb lb01 ip daddr set \

	symhash mod 1 map { 0 : 192.168.0.10 } fwd to "eth0"

This patch allows to use the symhash modulus with one element, in the same way that the other types of hashes and algorithms that uses the modulus parameter.

---
 net/netfilter/nft_hash.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/netfilter/nft_hash.c +1 -1
index 24f2f7567ddb7..010a565b40001 100644
@@ -131,7 +131,7 @@ static int nft_symhash_init(const struct nft_ctx *ctx,
 	priv->dreg = nft_parse_register(tb[NFTA_HASH_DREG]);
 
 	priv->modulus = ntohl(nla_get_be32(tb[NFTA_HASH_MODULUS]));
-	if (priv->modulus <= 1)
+	if (priv->modulus < 1)
 		return -ERANGE;
 
 	if (priv->offset + priv->modulus - 1 < priv->offset)
Signature · 1 line
2.20.1
author maintainer
#4 lore raw

PATCH 5/30 scripts/sphinx-pre-install: fix script for RHEL/CentOS

In reply to Sasha Levin

From: Mauro Carvalho Chehab <[email protected]>

[ Upstream commit b308467c916aa7acc5069802ab76a9f657434701 ]

There's a missing parenthesis at the script, with causes it to fail to detect non-Fedora releases (e. g. RHEL/CentOS).

Tested with Centos 7.6.1810.

---
 scripts/sphinx-pre-install | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
scripts/sphinx-pre-install +1 -1
index 067459760a7b0..3524dbc313163 100755
@@ -301,7 +301,7 @@ sub give_redhat_hints()
 	#
 	# Checks valid for RHEL/CentOS version 7.x.
 	#
-	if (! $system_release =~ /Fedora/) {
+	if (!($system_release =~ /Fedora/)) {
 		$map{"virtualenv"} = "python-virtualenv";
 	}
 
Signature · 1 line
2.20.1
author maintainer
#5 lore raw

PATCH 6/30 iscsi_ibft: make ISCSI_IBFT dependson ACPI instead of ISCSI_IBFT_FIND

In reply to Sasha Levin

From: Thomas Tai <[email protected]>

[ Upstream commit 94bccc34071094c165c79b515d21b63c78f7e968 ]

iscsi_ibft can use ACPI to find the iBFT entry during bootup, currently, ISCSI_IBFT depends on ISCSI_IBFT_FIND which is a X86 legacy way to find the iBFT by searching through the low memory. This patch changes the dependency so that other arch like ARM64 can use ISCSI_IBFT as long as the arch supports ACPI.

ibft_init() needs to use the global variable ibft_addr declared in iscsi_ibft_find.c. A #ifndef CONFIG_ISCSI_IBFT_FIND is needed to declare the variable if CONFIG_ISCSI_IBFT_FIND is not selected. Moving ibft_addr into the iscsi_ibft.c does not work because if ISCSI_IBFT is selected as a module, the arch/x86/kernel/setup.c won't be able to find the variable at compile time.

---
 drivers/firmware/Kconfig      | 5 +++--
 drivers/firmware/iscsi_ibft.c | 4 ++++
 2 files changed, 7 insertions(+), 2 deletions(-)
drivers/firmware/Kconfig +3 -2
index 6e4ed5a9c6fdc..42c4ff75281be 100644
@@ -156,7 +156,7 @@ config DMI_SCAN_MACHINE_NON_EFI_FALLBACK
 
 config ISCSI_IBFT_FIND
 	bool "iSCSI Boot Firmware Table Attributes"
-	depends on X86 && ACPI
+	depends on X86 && ISCSI_IBFT
 	default n
 	help
 	  This option enables the kernel to find the region of memory
@@ -167,7 +167,8 @@ config ISCSI_IBFT_FIND
 config ISCSI_IBFT
 	tristate "iSCSI Boot Firmware Table Attributes module"
 	select ISCSI_BOOT_SYSFS
-	depends on ISCSI_IBFT_FIND && SCSI && SCSI_LOWLEVEL
+	select ISCSI_IBFT_FIND if X86
+	depends on ACPI && SCSI && SCSI_LOWLEVEL
 	default	n
 	help
 	  This option enables support for detection and exposing of iSCSI
drivers/firmware/iscsi_ibft.c +4 -0
index 132b9bae4b6aa..220bbc91cebdb 100644
@@ -93,6 +93,10 @@ MODULE_DESCRIPTION("sysfs interface to BIOS iBFT information");
 MODULE_LICENSE("GPL");
 MODULE_VERSION(IBFT_ISCSI_VERSION);
 
+#ifndef CONFIG_ISCSI_IBFT_FIND
+struct acpi_table_ibft *ibft_addr;
+#endif
+
 struct ibft_hdr {
 	u8 id;
 	u8 version;
Signature · 1 line
2.20.1
author maintainer
#6 lore raw

PATCH 7/30 mac80211: don't warn about CW params when not using them

In reply to Sasha Levin

From: Brian Norris <[email protected]>

[ Upstream commit d2b3fe42bc629c2d4002f652b3abdfb2e72991c7 ]

ieee80211_set_wmm_default() normally sets up the initial CW min/max for each queue, except that it skips doing this if the driver doesn't support ->conf_tx. We still end up calling drv_conf_tx() in some cases (e.g., ieee80211_reconfig()), which also still won't do anything useful...except it complains here about the invalid CW parameters.

Let's just skip the WARN if we weren't going to do anything useful with the parameters.

---
 net/mac80211/driver-ops.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)
net/mac80211/driver-ops.c +9 -4
index bb886e7db47f1..f783d1377d9a8 100644
@@ -169,11 +169,16 @@ int drv_conf_tx(struct ieee80211_local *local,
 	if (!check_sdata_in_driver(sdata))
 		return -EIO;
 
-	if (WARN_ONCE(params->cw_min == 0 ||
-		      params->cw_min > params->cw_max,
-		      "%s: invalid CW_min/CW_max: %d/%d\n",
-		      sdata->name, params->cw_min, params->cw_max))
+	if (params->cw_min == 0 || params->cw_min > params->cw_max) {
+		/*
+		 * If we can't configure hardware anyway, don't warn. We may
+		 * never have initialized the CW parameters.
+		 */
+		WARN_ONCE(local->ops->conf_tx,
+			  "%s: invalid CW_min/CW_max: %d/%d\n",
+			  sdata->name, params->cw_min, params->cw_max);
 		return -EINVAL;
+	}
 
 	trace_drv_conf_tx(local, sdata, ac, params);
 	if (local->ops->conf_tx)
Signature · 1 line
2.20.1
author maintainer
#7 lore raw

PATCH 8/30 hwmon: (nct6775) Fix register address and added missed tolerance for nct6106

In reply to Sasha Levin

From: Björn Gerhart <[email protected]>

[ Upstream commit f3d43e2e45fd9d44ba52d20debd12cd4ee9c89bf ]

Fixed address of third NCT6106_REG_WEIGHT_DUTY_STEP, and added missed NCT6106_REG_TOLERANCE_H.

Fixes 6c009501ff200 ("hwmon: (nct6775) Add support for NCT6102D/6106D")
---
 drivers/hwmon/nct6775.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/hwmon/nct6775.c +2 -1
index ca9941fa741b7..7e14143ed1191 100644
@@ -769,7 +769,7 @@ static const u16 NCT6106_REG_TARGET[] = { 0x111, 0x121, 0x131 };
 static const u16 NCT6106_REG_WEIGHT_TEMP_SEL[] = { 0x168, 0x178, 0x188 };
 static const u16 NCT6106_REG_WEIGHT_TEMP_STEP[] = { 0x169, 0x179, 0x189 };
 static const u16 NCT6106_REG_WEIGHT_TEMP_STEP_TOL[] = { 0x16a, 0x17a, 0x18a };
-static const u16 NCT6106_REG_WEIGHT_DUTY_STEP[] = { 0x16b, 0x17b, 0x17c };
+static const u16 NCT6106_REG_WEIGHT_DUTY_STEP[] = { 0x16b, 0x17b, 0x18b };
 static const u16 NCT6106_REG_WEIGHT_TEMP_BASE[] = { 0x16c, 0x17c, 0x18c };
 static const u16 NCT6106_REG_WEIGHT_DUTY_BASE[] = { 0x16d, 0x17d, 0x18d };
 
@@ -3592,6 +3592,7 @@ static int nct6775_probe(struct platform_device *pdev)
 		data->REG_FAN_TIME[0] = NCT6106_REG_FAN_STOP_TIME;
 		data->REG_FAN_TIME[1] = NCT6106_REG_FAN_STEP_UP_TIME;
 		data->REG_FAN_TIME[2] = NCT6106_REG_FAN_STEP_DOWN_TIME;
+		data->REG_TOLERANCE_H = NCT6106_REG_TOLERANCE_H;
 		data->REG_PWM[0] = NCT6106_REG_PWM;
 		data->REG_PWM[1] = NCT6106_REG_FAN_START_OUTPUT;
 		data->REG_PWM[2] = NCT6106_REG_FAN_STOP_OUTPUT;
Signature · 1 line
2.20.1
author maintainer
#8 lore raw

PATCH 9/30 x86/mm: Check for pfn instead of page in vmalloc_sync_one()

In reply to Sasha Levin

From: Joerg Roedel <[email protected]>

[ Upstream commit 51b75b5b563a2637f9d8dc5bd02a31b2ff9e5ea0 ]

Do not require a struct page for the mapped memory location because it might not exist. This can happen when an ioremapped region is mapped with 2MB pages.

Fixes 5d72b4fba40ef ('x86, mm: support huge I/O mapping capability I/F')
---
 arch/x86/mm/fault.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/mm/fault.c +1 -1
index b162f92fd55cf..407522db3d9ea 100644
@@ -266,7 +266,7 @@ static inline pmd_t *vmalloc_sync_one(pgd_t *pgd, unsigned long address)
 	if (!pmd_present(*pmd))
 		set_pmd(pmd, *pmd_k);
 	else
-		BUG_ON(pmd_page(*pmd) != pmd_page(*pmd_k));
+		BUG_ON(pmd_pfn(*pmd) != pmd_pfn(*pmd_k));
 
 	return pmd_k;
 }
Signature · 1 line
2.20.1
author maintainer
#9 lore raw

PATCH 10/30 x86/mm: Sync also unmappings in vmalloc_sync_all()

In reply to Sasha Levin

From: Joerg Roedel <[email protected]>

[ Upstream commit 8e998fc24de47c55b47a887f6c95ab91acd4a720 ]

With huge-page ioremap areas the unmappings also need to be synced between all page-tables. Otherwise it can cause data corruption when a region is unmapped and later re-used.

Make the vmalloc_sync_one() function ready to sync unmappings and make sure vmalloc_sync_all() iterates over all page-tables even when an unmapped PMD is found.

Fixes 5d72b4fba40ef ('x86, mm: support huge I/O mapping capability I/F')
---
 arch/x86/mm/fault.c | 13 +++++--------
 1 file changed, 5 insertions(+), 8 deletions(-)
arch/x86/mm/fault.c +5 -8
index 407522db3d9ea..27cab342a0b25 100644
@@ -260,11 +260,12 @@ static inline pmd_t *vmalloc_sync_one(pgd_t *pgd, unsigned long address)
 
 	pmd = pmd_offset(pud, address);
 	pmd_k = pmd_offset(pud_k, address);
-	if (!pmd_present(*pmd_k))
-		return NULL;
 
-	if (!pmd_present(*pmd))
+	if (pmd_present(*pmd) != pmd_present(*pmd_k))
 		set_pmd(pmd, *pmd_k);
+
+	if (!pmd_present(*pmd_k))
+		return NULL;
 	else
 		BUG_ON(pmd_pfn(*pmd) != pmd_pfn(*pmd_k));
 
@@ -286,17 +287,13 @@ void vmalloc_sync_all(void)
 		spin_lock(&pgd_lock);
 		list_for_each_entry(page, &pgd_list, lru) {
 			spinlock_t *pgt_lock;
-			pmd_t *ret;
 
 			/* the pgt_lock only for Xen */
 			pgt_lock = &pgd_page_get_mm(page)->page_table_lock;
 
 			spin_lock(pgt_lock);
-			ret = vmalloc_sync_one(page_address(page), address);
+			vmalloc_sync_one(page_address(page), address);
 			spin_unlock(pgt_lock);
-
-			if (!ret)
-				break;
 		}
 		spin_unlock(&pgd_lock);
 	}
Signature · 1 line
2.20.1
author maintainer
#10 lore raw

PATCH 11/30 drm/msm: stop abusing dma_map/unmap for cache

In reply to Sasha Levin

From: Rob Clark <[email protected]>

[ Upstream commit 0036bc73ccbe7e600a3468bf8e8879b122252274 ]

Recently splats like this started showing up:

   WARNING: CPU: 4 PID: 251 at drivers/iommu/dma-iommu.c:451 __iommu_dma_unmap+0xb8/0xc0
   Modules linked in: ath10k_snoc ath10k_core fuse msm ath mac80211 uvcvideo cfg80211 videobuf2_vmalloc videobuf2_memops vide
   CPU: 4 PID: 251 Comm: kworker/u16:4 Tainted: G        W         5.2.0-rc5-next-20190619+ #2317
   Hardware name: LENOVO 81JL/LNVNB161216, BIOS 9UCN23WW(V1.06) 10/25/2018
   Workqueue: msm msm_gem_free_work [msm]
   pstate: 80c00005 (Nzcv daif +PAN +UAO)
   pc : __iommu_dma_unmap+0xb8/0xc0
   lr : __iommu_dma_unmap+0x54/0xc0
   sp : ffff0000119abce0
   x29: ffff0000119abce0 x28: 0000000000000000
   x27: ffff8001f9946648 x26: ffff8001ec271068
   x25: 0000000000000000 x24: ffff8001ea3580a8
   x23: ffff8001f95ba010 x22: ffff80018e83ba88
   x21: ffff8001e548f000 x20: fffffffffffff000
   x19: 0000000000001000 x18: 00000000c00001fe
   x17: 0000000000000000 x16: 0000000000000000
   x15: ffff000015b70068 x14: 0000000000000005
   x13: 0003142cc1be1768 x12: 0000000000000001
   x11: ffff8001f6de9100 x10: 0000000000000009
   x9 : ffff000015b78000 x8 : 0000000000000000
   x7 : 0000000000000001 x6 : fffffffffffff000
   x5 : 0000000000000fff x4 : ffff00001065dbc8
   x3 : 000000000000000d x2 : 0000000000001000
   x1 : fffffffffffff000 x0 : 0000000000000000
   Call trace:
    __iommu_dma_unmap+0xb8/0xc0
    iommu_dma_unmap_sg+0x98/0xb8
    put_pages+0x5c/0xf0 [msm]
    msm_gem_free_work+0x10c/0x150 [msm]
    process_one_work+0x1e0/0x330
    worker_thread+0x40/0x438
    kthread+0x12c/0x130
    ret_from_fork+0x10/0x18
   ---[ end trace afc0dc5ab81a06bf ]---

Not quite sure what triggered that, but we really shouldn't be abusing dma_{map,unmap}_sg() for cache maint.

---
 drivers/gpu/drm/msm/msm_gem.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/msm/msm_gem.c +2 -2
index f2df718af370d..3a91ccd92c473 100644
@@ -108,7 +108,7 @@ static struct page **get_pages(struct drm_gem_object *obj)
 		 * because display controller, GPU, etc. are not coherent:
 		 */
 		if (msm_obj->flags & (MSM_BO_WC|MSM_BO_UNCACHED))
-			dma_map_sg(dev->dev, msm_obj->sgt->sgl,
+			dma_sync_sg_for_device(dev->dev, msm_obj->sgt->sgl,
 					msm_obj->sgt->nents, DMA_BIDIRECTIONAL);
 	}
 
@@ -138,7 +138,7 @@ static void put_pages(struct drm_gem_object *obj)
 			 * GPU, etc. are not coherent:
 			 */
 			if (msm_obj->flags & (MSM_BO_WC|MSM_BO_UNCACHED))
-				dma_unmap_sg(obj->dev->dev, msm_obj->sgt->sgl,
+				dma_sync_sg_for_cpu(obj->dev->dev, msm_obj->sgt->sgl,
 					     msm_obj->sgt->nents,
 					     DMA_BIDIRECTIONAL);
 
Signature · 1 line
2.20.1
#11 lore raw

Re: [Freedreno] [PATCH AUTOSEL 4.14 11/30] drm/msm: stop abusing dma_map/unmap for cache

Quoted 5 lines · Sasha Levin

From: Rob Clark <[email protected]>

[ Upstream commit 0036bc73ccbe7e600a3468bf8e8879b122252274 ]

Hi Sasha,
FYI, this commit should be paired with
https://cgit.freedesktop.org/drm/drm/commit/?h=drm-fixes&id=3de433c5b38af49a5fc7602721e2ab5d39f1e69c

Sean

Quoted 80 lines

Recently splats like this started showing up:

   WARNING: CPU: 4 PID: 251 at drivers/iommu/dma-iommu.c:451 __iommu_dma_unmap+0xb8/0xc0
   Modules linked in: ath10k_snoc ath10k_core fuse msm ath mac80211 uvcvideo cfg80211 videobuf2_vmalloc videobuf2_memops vide
   CPU: 4 PID: 251 Comm: kworker/u16:4 Tainted: G        W         5.2.0-rc5-next-20190619+ #2317
   Hardware name: LENOVO 81JL/LNVNB161216, BIOS 9UCN23WW(V1.06) 10/25/2018
   Workqueue: msm msm_gem_free_work [msm]
   pstate: 80c00005 (Nzcv daif +PAN +UAO)
   pc : __iommu_dma_unmap+0xb8/0xc0
   lr : __iommu_dma_unmap+0x54/0xc0
   sp : ffff0000119abce0
   x29: ffff0000119abce0 x28: 0000000000000000
   x27: ffff8001f9946648 x26: ffff8001ec271068
   x25: 0000000000000000 x24: ffff8001ea3580a8
   x23: ffff8001f95ba010 x22: ffff80018e83ba88
   x21: ffff8001e548f000 x20: fffffffffffff000
   x19: 0000000000001000 x18: 00000000c00001fe
   x17: 0000000000000000 x16: 0000000000000000
   x15: ffff000015b70068 x14: 0000000000000005
   x13: 0003142cc1be1768 x12: 0000000000000001
   x11: ffff8001f6de9100 x10: 0000000000000009
   x9 : ffff000015b78000 x8 : 0000000000000000
   x7 : 0000000000000001 x6 : fffffffffffff000
   x5 : 0000000000000fff x4 : ffff00001065dbc8
   x3 : 000000000000000d x2 : 0000000000001000
   x1 : fffffffffffff000 x0 : 0000000000000000
   Call trace:
    __iommu_dma_unmap+0xb8/0xc0
    iommu_dma_unmap_sg+0x98/0xb8
    put_pages+0x5c/0xf0 [msm]
    msm_gem_free_work+0x10c/0x150 [msm]
    process_one_work+0x1e0/0x330
    worker_thread+0x40/0x438
    kthread+0x12c/0x130
    ret_from_fork+0x10/0x18
   ---[ end trace afc0dc5ab81a06bf ]---

Not quite sure what triggered that, but we really shouldn't be abusing dma_{map,unmap}_sg() for cache maint.

---
 drivers/gpu/drm/msm/msm_gem.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/msm/msm_gem.c +2 -3
index f2df718af370d..3a91ccd92c473 100644
@@ -108,7 +108,7 @@ static struct page **get_pages(struct drm_gem_object *obj)
                 * because display controller, GPU, etc. are not coherent:
                 */
                if (msm_obj->flags & (MSM_BO_WC|MSM_BO_UNCACHED))
-                       dma_map_sg(dev->dev, msm_obj->sgt->sgl,
+                       dma_sync_sg_for_device(dev->dev, msm_obj->sgt->sgl,
                                        msm_obj->sgt->nents, DMA_BIDIRECTIONAL);
        }
 
@@ -138,7 +138,7 @@ static void put_pages(struct drm_gem_object *obj)
                         * GPU, etc. are not coherent:
                         */
                        if (msm_obj->flags & (MSM_BO_WC|MSM_BO_UNCACHED))
-                               dma_unmap_sg(obj->dev->dev, msm_obj->sgt->sgl,
+                               dma_sync_sg_for_cpu(obj->dev->dev, msm_obj->sgt->sgl,
                                             msm_obj->sgt->nents,
                                             DMA_BIDIRECTIONAL);
 
--

2.20.1

_______________________________________________
Freedreno mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/freedreno

author maintainer
#12 lore raw

PATCH 12/30 drm: silence variable 'conn' set but not used

In reply to Sasha Levin

From: Qian Cai <[email protected]>

[ Upstream commit bbb6fc43f131f77fcb7ae8081f6d7c51396a2120 ]

The "struct drm_connector" iteration cursor from "for_each_new_connector_in_state" is never used in atomic_remove_fb() which generates a compilation warning,

drivers/gpu/drm/drm_framebuffer.c: In function 'atomic_remove_fb':
drivers/gpu/drm/drm_framebuffer.c:838:24: warning: variable 'conn' set but not used [-Wunused-but-set-variable]

Silence it by marking "conn" __maybe_unused.

---
 drivers/gpu/drm/drm_framebuffer.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/drm_framebuffer.c +1 -1
index c21e10c780ac5..af40189cdb60c 100644
@@ -773,7 +773,7 @@ static int atomic_remove_fb(struct drm_framebuffer *fb)
 	struct drm_device *dev = fb->dev;
 	struct drm_atomic_state *state;
 	struct drm_plane *plane;
-	struct drm_connector *conn;
+	struct drm_connector *conn __maybe_unused;
 	struct drm_connector_state *conn_state;
 	int i, ret = 0;
 	unsigned plane_mask;
Signature · 6 lines
2.20.1

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

PATCH 13/30 cpufreq/pasemi: fix use-after-free in pas_cpufreq_cpu_init()

In reply to Sasha Levin

From: Wen Yang <[email protected]>

[ Upstream commit e0a12445d1cb186d875410d093a00d215bec6a89 ]

The cpu variable is still being used in the of_get_property() call after the of_node_put() call, which may result in use-after-free.

Fixes a9acc26b75f6 ("cpufreq/pasemi: fix possible object reference leak")
---
 drivers/cpufreq/pasemi-cpufreq.c | 23 +++++++++--------------
 1 file changed, 9 insertions(+), 14 deletions(-)
drivers/cpufreq/pasemi-cpufreq.c +9 -14
index 8456492124f0c..d1bdd8f622476 100644
@@ -145,10 +145,18 @@ static int pas_cpufreq_cpu_init(struct cpufreq_policy *policy)
 	int err = -ENODEV;
 
 	cpu = of_get_cpu_node(policy->cpu, NULL);
+	if (!cpu)
+		goto out;
 
+	max_freqp = of_get_property(cpu, "clock-frequency", NULL);
 	of_node_put(cpu);
-	if (!cpu)
+	if (!max_freqp) {
+		err = -EINVAL;
 		goto out;
+	}
+
+	/* we need the freq in kHz */
+	max_freq = *max_freqp / 1000;
 
 	dn = of_find_compatible_node(NULL, NULL, "1682m-sdc");
 	if (!dn)
@@ -185,16 +193,6 @@ static int pas_cpufreq_cpu_init(struct cpufreq_policy *policy)
 	}
 
 	pr_debug("init cpufreq on CPU %d\n", policy->cpu);
-
-	max_freqp = of_get_property(cpu, "clock-frequency", NULL);
-	if (!max_freqp) {
-		err = -EINVAL;
-		goto out_unmap_sdcpwr;
-	}
-
-	/* we need the freq in kHz */
-	max_freq = *max_freqp / 1000;
-
 	pr_debug("max clock-frequency is at %u kHz\n", max_freq);
 	pr_debug("initializing frequency table\n");
 
@@ -212,9 +210,6 @@ static int pas_cpufreq_cpu_init(struct cpufreq_policy *policy)
 
 	return cpufreq_generic_init(policy, pas_freqs, get_gizmo_latency());
 
-out_unmap_sdcpwr:
-	iounmap(sdcpwr_mapbase);
-
 out_unmap_sdcasr:
 	iounmap(sdcasr_mapbase);
 out:
Signature · 1 line
2.20.1
author maintainer
#14 lore raw

PATCH 14/30 s390/qdio: add sanity checks to the fast-requeue path

In reply to Sasha Levin

From: Julian Wiedmann <[email protected]>

[ Upstream commit a6ec414a4dd529eeac5c3ea51c661daba3397108 ]

If the device driver were to send out a full queue's worth of SBALs, current code would end up discovering the last of those SBALs as PRIMED and erroneously skip the SIGA-w. This immediately stalls the queue.

Add a check to not attempt fast-requeue in this case. While at it also make sure that the state of the previous SBAL was successfully extracted before inspecting it.

---
 drivers/s390/cio/qdio_main.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
drivers/s390/cio/qdio_main.c +6 -6
index ab8dd81fbc2b1..1a40c73961b83 100644
@@ -1577,13 +1577,13 @@ static int handle_outbound(struct qdio_q *q, unsigned int callflags,
 		rc = qdio_kick_outbound_q(q, phys_aob);
 	} else if (need_siga_sync(q)) {
 		rc = qdio_siga_sync_q(q);
+	} else if (count < QDIO_MAX_BUFFERS_PER_Q &&
+		   get_buf_state(q, prev_buf(bufnr), &state, 0) > 0 &&
+		   state == SLSB_CU_OUTPUT_PRIMED) {
+		/* The previous buffer is not processed yet, tack on. */
+		qperf_inc(q, fast_requeue);
 	} else {
-		/* try to fast requeue buffers */
-		get_buf_state(q, prev_buf(bufnr), &state, 0);
-		if (state != SLSB_CU_OUTPUT_PRIMED)
-			rc = qdio_kick_outbound_q(q, 0);
-		else
-			qperf_inc(q, fast_requeue);
+		rc = qdio_kick_outbound_q(q, 0);
 	}
 
 	/* in case of SIGA errors we must process the error immediately */
Signature · 1 line
2.20.1
author maintainer
#15 lore raw

PATCH 15/30 ALSA: compress: Fix regression on compressed capture streams

In reply to Sasha Levin

From: Charles Keepax <[email protected]>

[ Upstream commit 4475f8c4ab7b248991a60d9c02808dbb813d6be8 ]

A previous fix to the stop handling on compressed capture streams causes some knock on issues. The previous fix updated snd_compr_drain_notify to set the state back to PREPARED for capture streams. This causes some issues however as the handling for snd_compr_poll differs between the two states and some user-space applications were relying on the poll failing after the stream had been stopped.

To correct this regression whilst still fixing the original problem the patch was addressing, update the capture handling to skip the PREPARED state rather than skipping the SETUP state as it has done until now.

Fixes 4f2ab5e1d13d ("ALSA: compress: Fix stop handling on compressed capture streams")
---
 include/sound/compress_driver.h |  5 +----
 sound/core/compress_offload.c   | 16 +++++++++++-----
 2 files changed, 12 insertions(+), 9 deletions(-)
include/sound/compress_driver.h +1 -4
index 392bac18398ba..33a07c3badf01 100644
@@ -186,10 +186,7 @@ static inline void snd_compr_drain_notify(struct snd_compr_stream *stream)
 	if (snd_BUG_ON(!stream))
 		return;
 
-	if (stream->direction == SND_COMPRESS_PLAYBACK)
-		stream->runtime->state = SNDRV_PCM_STATE_SETUP;
-	else
-		stream->runtime->state = SNDRV_PCM_STATE_PREPARED;
+	stream->runtime->state = SNDRV_PCM_STATE_SETUP;
 
 	wake_up(&stream->runtime->sleep);
 }
sound/core/compress_offload.c +11 -5
index 555df64d46ffc..cf1317546b0ff 100644
@@ -575,10 +575,7 @@ snd_compr_set_params(struct snd_compr_stream *stream, unsigned long arg)
 		stream->metadata_set = false;
 		stream->next_track = false;
 
-		if (stream->direction == SND_COMPRESS_PLAYBACK)
-			stream->runtime->state = SNDRV_PCM_STATE_SETUP;
-		else
-			stream->runtime->state = SNDRV_PCM_STATE_PREPARED;
+		stream->runtime->state = SNDRV_PCM_STATE_SETUP;
 	} else {
 		return -EPERM;
 	}
@@ -694,8 +691,17 @@ static int snd_compr_start(struct snd_compr_stream *stream)
 {
 	int retval;
 
-	if (stream->runtime->state != SNDRV_PCM_STATE_PREPARED)
+	switch (stream->runtime->state) {
+	case SNDRV_PCM_STATE_SETUP:
+		if (stream->direction != SND_COMPRESS_CAPTURE)
+			return -EPERM;
+		break;
+	case SNDRV_PCM_STATE_PREPARED:
+		break;
+	default:
 		return -EPERM;
+	}
+
 	retval = stream->ops->trigger(stream, SNDRV_PCM_TRIGGER_START);
 	if (!retval)
 		stream->runtime->state = SNDRV_PCM_STATE_RUNNING;
Signature · 1 line
2.20.1
author maintainer
#16 lore raw

PATCH 16/30 ALSA: compress: Prevent bypasses of set_params

In reply to Sasha Levin

From: Charles Keepax <[email protected]>

[ Upstream commit 26c3f1542f5064310ad26794c09321780d00c57d ]

Currently, whilst in SNDRV_PCM_STATE_OPEN it is possible to call snd_compr_stop, snd_compr_drain and snd_compr_partial_drain, which allow a transition to SNDRV_PCM_STATE_SETUP. The stream should only be able to move to the setup state once it has received a SNDRV_COMPRESS_SET_PARAMS ioctl. Fix this issue by not allowing those ioctls whilst in the open state.

---
 sound/core/compress_offload.c | 30 ++++++++++++++++++++++++------
 1 file changed, 24 insertions(+), 6 deletions(-)
sound/core/compress_offload.c +24 -6
index cf1317546b0ff..1538fbc7562b8 100644
@@ -712,9 +712,15 @@ static int snd_compr_stop(struct snd_compr_stream *stream)
 {
 	int retval;
 
-	if (stream->runtime->state == SNDRV_PCM_STATE_PREPARED ||
-			stream->runtime->state == SNDRV_PCM_STATE_SETUP)
+	switch (stream->runtime->state) {
+	case SNDRV_PCM_STATE_OPEN:
+	case SNDRV_PCM_STATE_SETUP:
+	case SNDRV_PCM_STATE_PREPARED:
 		return -EPERM;
+	default:
+		break;
+	}
+
 	retval = stream->ops->trigger(stream, SNDRV_PCM_TRIGGER_STOP);
 	if (!retval) {
 		snd_compr_drain_notify(stream);
@@ -802,9 +808,14 @@ static int snd_compr_drain(struct snd_compr_stream *stream)
 {
 	int retval;
 
-	if (stream->runtime->state == SNDRV_PCM_STATE_PREPARED ||
-			stream->runtime->state == SNDRV_PCM_STATE_SETUP)
+	switch (stream->runtime->state) {
+	case SNDRV_PCM_STATE_OPEN:
+	case SNDRV_PCM_STATE_SETUP:
+	case SNDRV_PCM_STATE_PREPARED:
 		return -EPERM;
+	default:
+		break;
+	}
 
 	retval = stream->ops->trigger(stream, SND_COMPR_TRIGGER_DRAIN);
 	if (retval) {
@@ -841,9 +852,16 @@ static int snd_compr_next_track(struct snd_compr_stream *stream)
 static int snd_compr_partial_drain(struct snd_compr_stream *stream)
 {
 	int retval;
-	if (stream->runtime->state == SNDRV_PCM_STATE_PREPARED ||
-			stream->runtime->state == SNDRV_PCM_STATE_SETUP)
+
+	switch (stream->runtime->state) {
+	case SNDRV_PCM_STATE_OPEN:
+	case SNDRV_PCM_STATE_SETUP:
+	case SNDRV_PCM_STATE_PREPARED:
 		return -EPERM;
+	default:
+		break;
+	}
+
 	/* stream can be drained only when next track has been signalled */
 	if (stream->next_track == false)
 		return -EPERM;
Signature · 1 line
2.20.1
author maintainer
#17 lore raw

PATCH 17/30 ALSA: compress: Don't allow paritial drain operations on capture streams

In reply to Sasha Levin

From: Charles Keepax <[email protected]>

[ Upstream commit a70ab8a8645083f3700814e757f2940a88b7ef88 ]

Partial drain and next track are intended for gapless playback and don't really have an obvious interpretation for a capture stream, so makes sense to not allow those operations on capture streams.

---
 sound/core/compress_offload.c | 8 ++++++++
 1 file changed, 8 insertions(+)
sound/core/compress_offload.c +8 -0
index 1538fbc7562b8..b4f1536b17cb5 100644
@@ -835,6 +835,10 @@ static int snd_compr_next_track(struct snd_compr_stream *stream)
 	if (stream->runtime->state != SNDRV_PCM_STATE_RUNNING)
 		return -EPERM;
 
+	/* next track doesn't have any meaning for capture streams */
+	if (stream->direction == SND_COMPRESS_CAPTURE)
+		return -EPERM;
+
 	/* you can signal next track if this is intended to be a gapless stream
 	 * and current track metadata is set
 	 */
@@ -862,6 +866,10 @@ static int snd_compr_partial_drain(struct snd_compr_stream *stream)
 		break;
 	}
 
+	/* partial drain doesn't have any meaning for capture streams */
+	if (stream->direction == SND_COMPRESS_CAPTURE)
+		return -EPERM;
+
 	/* stream can be drained only when next track has been signalled */
 	if (stream->next_track == false)
 		return -EPERM;
Signature · 1 line
2.20.1
author maintainer
#18 lore raw

PATCH 18/30 ALSA: compress: Be more restrictive about when a drain is allowed

In reply to Sasha Levin

From: Charles Keepax <[email protected]>

[ Upstream commit 3b8179944cb0dd53e5223996966746cdc8a60657 ]

Draining makes little sense in the situation of hardware overrun, as the hardware will have consumed all its available samples. Additionally, draining whilst the stream is paused would presumably get stuck as no data is being consumed on the DSP side.

---
 sound/core/compress_offload.c | 6 ++++++
 1 file changed, 6 insertions(+)
sound/core/compress_offload.c +6 -0
index b4f1536b17cb5..2e2d184684911 100644
@@ -812,7 +812,10 @@ static int snd_compr_drain(struct snd_compr_stream *stream)
 	case SNDRV_PCM_STATE_OPEN:
 	case SNDRV_PCM_STATE_SETUP:
 	case SNDRV_PCM_STATE_PREPARED:
+	case SNDRV_PCM_STATE_PAUSED:
 		return -EPERM;
+	case SNDRV_PCM_STATE_XRUN:
+		return -EPIPE;
 	default:
 		break;
 	}
@@ -861,7 +864,10 @@ static int snd_compr_partial_drain(struct snd_compr_stream *stream)
 	case SNDRV_PCM_STATE_OPEN:
 	case SNDRV_PCM_STATE_SETUP:
 	case SNDRV_PCM_STATE_PREPARED:
+	case SNDRV_PCM_STATE_PAUSED:
 		return -EPERM;
+	case SNDRV_PCM_STATE_XRUN:
+		return -EPIPE;
 	default:
 		break;
 	}
Signature · 1 line
2.20.1
author maintainer
#19 lore raw

PATCH 19/30 perf tools: Fix proper buffer size for feature processing

In reply to Sasha Levin

From: Jiri Olsa <[email protected]>

[ Upstream commit 79b2fe5e756163897175a8f57d66b26cd9befd59 ]

After Song Liu's segfault fix for pipe mode, Arnaldo reported following error:

  # perf record -o - | perf script
  0x514 [0x1ac]: failed to process type: 80

It's caused by wrong buffer size setup in feature processing, which makes cpu topology feature fail, because it's using buffer size to recognize its header version.

Cc Alexander Shishkin <[email protected]>
Cc David Carrillo-Cisneros <[email protected]>
Cc Kan Liang <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Song Liu <[email protected]>
Fixes e9def1b2e74e ("perf tools: Add feature header record to pipe-mode")
---
 tools/perf/util/header.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/header.c +1 -1
index 26437143c9406..c892a28e7b048 100644
@@ -3081,7 +3081,7 @@ int perf_event__process_feature(struct perf_tool *tool,
 		return 0;
 
 	ff.buf  = (void *)fe->data;
-	ff.size = event->header.size - sizeof(event->header);
+	ff.size = event->header.size - sizeof(*fe);
 	ff.ph = &session->header;
 
 	if (feat_ops[feat].process(&ff, NULL))
Signature · 1 line
2.20.1
author maintainer
#20 lore raw

PATCH 20/30 perf probe: Avoid calling freeing routine multiple times for same pointer

In reply to Sasha Levin

From: Arnaldo Carvalho de Melo <[email protected]>

[ Upstream commit d95daf5accf4a72005daa13fbb1d1bd8709f2861 ]

When perf_add_probe_events() we call cleanup_perf_probe_events() for the pev pointer it receives, then, as part of handling this failure the main 'perf probe' goes on and calls cleanup_params() and that will again call cleanup_perf_probe_events()for the same pointer, so just set nevents to zero when handling the failure of perf_add_probe_events() to avoid the double free.

---
 tools/perf/builtin-probe.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
tools/perf/builtin-probe.c +10 -0
index c0065923a525f..e1ac51aaedcff 100644
@@ -712,6 +712,16 @@ __cmd_probe(int argc, const char **argv)
 
 		ret = perf_add_probe_events(params.events, params.nevents);
 		if (ret < 0) {
+
+			/*
+			 * When perf_add_probe_events() fails it calls
+			 * cleanup_perf_probe_events(pevs, npevs), i.e.
+			 * cleanup_perf_probe_events(params.events, params.nevents), which
+			 * will call clear_perf_probe_event(), so set nevents to zero
+			 * to avoid cleanup_params() to call clear_perf_probe_event() again
+			 * on the same pevs.
+			 */
+			params.nevents = 0;
 			pr_err_with_code("  Error: Failed to add events.", ret);
 			return ret;
 		}
Signature · 1 line
2.20.1
author maintainer
#21 lore raw

PATCH 21/30 drbd: dynamically allocate shash descriptor

In reply to Sasha Levin

From: Arnd Bergmann <[email protected]>

[ Upstream commit 77ce56e2bfaa64127ae5e23ef136c0168b818777 ]

Building with clang and KASAN, we get a warning about an overly large stack frame on 32-bit architectures:

drivers/block/drbd/drbd_receiver.c:921:31: error: stack frame size of 1280 bytes in function 'conn_connect'

      [-Werror,-Wframe-larger-than=]

We already allocate other data dynamically in this function, so just do the same for the shash descriptor, which makes up most of this memory.

---
 drivers/block/drbd/drbd_receiver.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)
drivers/block/drbd/drbd_receiver.c +12 -2
index 1aad373da50e2..8fbdfaacc2226 100644
@@ -5237,7 +5237,7 @@ static int drbd_do_auth(struct drbd_connection *connection)
 	unsigned int key_len;
 	char secret[SHARED_SECRET_MAX]; /* 64 byte */
 	unsigned int resp_size;
-	SHASH_DESC_ON_STACK(desc, connection->cram_hmac_tfm);
+	struct shash_desc *desc;
 	struct packet_info pi;
 	struct net_conf *nc;
 	int err, rv;
@@ -5250,6 +5250,13 @@ static int drbd_do_auth(struct drbd_connection *connection)
 	memcpy(secret, nc->shared_secret, key_len);
 	rcu_read_unlock();
 
+	desc = kmalloc(sizeof(struct shash_desc) +
+		       crypto_shash_descsize(connection->cram_hmac_tfm),
+		       GFP_KERNEL);
+	if (!desc) {
+		rv = -1;
+		goto fail;
+	}
 	desc->tfm = connection->cram_hmac_tfm;
 	desc->flags = 0;
 
@@ -5392,7 +5399,10 @@ static int drbd_do_auth(struct drbd_connection *connection)
 	kfree(peers_ch);
 	kfree(response);
 	kfree(right_response);
-	shash_desc_zero(desc);
+	if (desc) {
+		shash_desc_zero(desc);
+		kfree(desc);
+	}
 
 	return rv;
 }
Signature · 1 line
2.20.1
author maintainer
#22 lore raw

PATCH 22/30 ACPI/IORT: Fix off-by-one check in iort_dev_find_its_id()

In reply to Sasha Levin

From: Lorenzo Pieralisi <[email protected]>

[ Upstream commit 5a46d3f71d5e5a9f82eabc682f996f1281705ac7 ]

Static analysis identified that index comparison against ITS entries in iort_dev_find_its_id() is off by one.

Update the comparison condition and clarify the resulting error message.

Fixes 4bf2efd26d76 ("ACPI: Add new IORT functions to support MSI domain handling")
Cc Dan Carpenter <[email protected]>
Cc Will Deacon <[email protected]>
Cc Hanjun Guo <[email protected]>
Cc Sudeep Holla <[email protected]>
Cc Catalin Marinas <[email protected]>
Cc Robin Murphy <[email protected]>
---
 drivers/acpi/arm64/iort.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/acpi/arm64/iort.c +2 -2
index ca414910710ea..b0a7afd4e7d35 100644
@@ -506,8 +506,8 @@ static int iort_dev_find_its_id(struct device *dev, u32 req_id,
 
 	/* Move to ITS specific data */
 	its = (struct acpi_iort_its_group *)node->node_data;
-	if (idx > its->its_count) {
-		dev_err(dev, "requested ITS ID index [%d] is greater than available [%d]\n",
+	if (idx >= its->its_count) {
+		dev_err(dev, "requested ITS ID index [%d] overruns ITS entries [%d]\n",
 			idx, its->its_count);
 		return -ENXIO;
 	}
Signature · 1 line
2.20.1
author maintainer
#23 lore raw

PATCH 23/30 ARM: davinci: fix sleep.S build error on ARMv4

In reply to Sasha Levin

From: Arnd Bergmann <[email protected]>

[ Upstream commit d64b212ea960db4276a1d8372bd98cb861dfcbb0 ]

When building a multiplatform kernel that includes armv4 support, the default target CPU does not support the blx instruction, which leads to a build failure:

arch/arm/mach-davinci/sleep.S: Assembler messages:
arch/arm/mach-davinci/sleep.S:56: Error: selected processor does not support `blx ip' in ARM mode

Add a .arch statement in the sources to make this file build.

---
 arch/arm/mach-davinci/sleep.S | 1 +
 1 file changed, 1 insertion(+)
arch/arm/mach-davinci/sleep.S +1 -0
index cd350dee4df37..efcd400b2abb3 100644
@@ -37,6 +37,7 @@
 #define DEEPSLEEP_SLEEPENABLE_BIT	BIT(31)
 
 	.text
+	.arch	armv5te
 /*
  * Move DaVinci into deep sleep state
  *
Signature · 1 line
2.20.1
author maintainer
#24 lore raw

PATCH 24/30 scsi: megaraid_sas: fix panic on loading firmware crashdump

In reply to Sasha Levin

From: Junxiao Bi <[email protected]>

[ Upstream commit 3b5f307ef3cb5022bfe3c8ca5b8f2114d5bf6c29 ]

While loading fw crashdump in function fw_crash_buffer_show(), left bytes in one dma chunk was not checked, if copying size over it, overflow access will cause kernel panic.

---
 drivers/scsi/megaraid/megaraid_sas_base.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/scsi/megaraid/megaraid_sas_base.c +3 -0
index 73acd3e9ded75..8595d83229b77 100644
@@ -2976,6 +2976,7 @@ megasas_fw_crash_buffer_show(struct device *cdev,
 	u32 size;
 	unsigned long buff_addr;
 	unsigned long dmachunk = CRASH_DMA_BUF_SIZE;
+	unsigned long chunk_left_bytes;
 	unsigned long src_addr;
 	unsigned long flags;
 	u32 buff_offset;
@@ -3001,6 +3002,8 @@ megasas_fw_crash_buffer_show(struct device *cdev,
 	}
 
 	size = (instance->fw_crash_buffer_size * dmachunk) - buff_offset;
+	chunk_left_bytes = dmachunk - (buff_offset % dmachunk);
+	size = (size > chunk_left_bytes) ? chunk_left_bytes : size;
 	size = (size >= PAGE_SIZE) ? (PAGE_SIZE - 1) : size;
 
 	src_addr = (unsigned long)instance->crash_buf[buff_offset / dmachunk] +
Signature · 1 line
2.20.1
author maintainer
#25 lore raw

PATCH 25/30 scsi: ibmvfc: fix WARN_ON during event pool release

In reply to Sasha Levin

From: Tyrel Datwyler <[email protected]>

[ Upstream commit 5578257ca0e21056821e6481bd534ba267b84e58 ]

While removing an ibmvfc client adapter a WARN_ON like the following WARN_ON is seen in the kernel log:

WARNING: CPU: 6 PID: 5421 at ./include/linux/dma-mapping.h:541
ibmvfc_free_event_pool+0x12c/0x1f0 [ibmvfc]
CPU: 6 PID: 5421 Comm: rmmod Tainted: G            E     4.17.0-rc1-next-20180419-autotest #1
NIP:  d00000000290328c LR: d00000000290325c CTR: c00000000036ee20
REGS: c000000288d1b7e0 TRAP: 0700   Tainted: G            E      (4.17.0-rc1-next-20180419-autotest)
MSR:  800000010282b033 <SF,VEC,VSX,EE,FP,ME,IR,DR,RI,LE,TM[E]>  CR: 44008828  XER: 20000000

CFAR: c00000000036e408 SOFTE: 1
GPR00: d00000000290325c c000000288d1ba60 d000000002917900 c000000289d75448
GPR04: 0000000000000071 c0000000ff870000 0000000018040000 0000000000000001
GPR08: 0000000000000000 c00000000156e838 0000000000000001 d00000000290c640
GPR12: c00000000036ee20 c00000001ec4dc00 0000000000000000 0000000000000000
GPR16: 0000000000000000 0000000000000000 00000100276901e0 0000000010020598
GPR20: 0000000010020550 0000000010020538 0000000010020578 00000000100205b0
GPR24: 0000000000000000 0000000000000000 0000000010020590 5deadbeef0000100
GPR28: 5deadbeef0000200 d000000002910b00 0000000000000071 c0000002822f87d8 NIP [d00000000290328c] ibmvfc_free_event_pool+0x12c/0x1f0 [ibmvfc] LR [d00000000290325c] ibmvfc_free_event_pool+0xfc/0x1f0 [ibmvfc]

Call Trace:
[c000000288d1ba60] [d00000000290325c] ibmvfc_free_event_pool+0xfc/0x1f0 [ibmvfc] (unreliable)
[c000000288d1baf0] [d000000002909390] ibmvfc_abort_task_set+0x7b0/0x8b0 [ibmvfc]
[c000000288d1bb70] [c0000000000d8c68] vio_bus_remove+0x68/0x100
[c000000288d1bbb0] [c0000000007da7c4] device_release_driver_internal+0x1f4/0x2d0
[c000000288d1bc00] [c0000000007da95c] driver_detach+0x7c/0x100
[c000000288d1bc40] [c0000000007d8af4] bus_remove_driver+0x84/0x140
[c000000288d1bcb0] [c0000000007db6ac] driver_unregister+0x4c/0xa0
[c000000288d1bd20] [c0000000000d6e7c] vio_unregister_driver+0x2c/0x50
[c000000288d1bd50] [d00000000290ba0c] cleanup_module+0x24/0x15e0 [ibmvfc]
[c000000288d1bd70] [c0000000001dadb0] sys_delete_module+0x220/0x2d0
[c000000288d1be30] [c00000000000b284] system_call+0x58/0x6c

Instruction dump:
e8410018 e87f0068 809f0078 e8bf0080 e8df0088 2fa30000 419e008c e9230200 2fa90000 419e0080 894d098a 794a07e0 <0b0a0000> e9290008 2fa90000 419e0028

This is tripped as a result of irqs being disabled during the call to dma_free_coherent() by ibmvfc_free_event_pool(). At this point in the code path we have quiesced the adapter and its overly paranoid anyways to be holding the host lock.

---
 drivers/scsi/ibmvscsi/ibmvfc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/scsi/ibmvscsi/ibmvfc.c +1 -1
index a06b24a61622c..34612add3829f 100644
@@ -4876,8 +4876,8 @@ static int ibmvfc_remove(struct vio_dev *vdev)
 
 	spin_lock_irqsave(vhost->host->host_lock, flags);
 	ibmvfc_purge_requests(vhost, DID_ERROR);
-	ibmvfc_free_event_pool(vhost);
 	spin_unlock_irqrestore(vhost->host->host_lock, flags);
+	ibmvfc_free_event_pool(vhost);
 
 	ibmvfc_free_mem(vhost);
 	spin_lock(&ibmvfc_driver_lock);
Signature · 1 line
2.20.1
author maintainer
#26 lore raw

PATCH 26/30 scsi: scsi_dh_alua: always use a 2 second delay before retrying RTPG

In reply to Sasha Levin

From: Hannes Reinecke <[email protected]>

[ Upstream commit 20122994e38aef0ae50555884d287adde6641c94 ]

Retrying immediately after we've received a 'transitioning' sense code is pretty much pointless, we should always use a delay before retrying. So ensure the default delay is applied before retrying.

---
 drivers/scsi/device_handler/scsi_dh_alua.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/scsi/device_handler/scsi_dh_alua.c +6 -1
index 09c6a16fab93f..41f5f64101630 100644
@@ -53,6 +53,7 @@
 #define ALUA_FAILOVER_TIMEOUT		60
 #define ALUA_FAILOVER_RETRIES		5
 #define ALUA_RTPG_DELAY_MSECS		5
+#define ALUA_RTPG_RETRY_DELAY		2
 
 /* device handler flags */
 #define ALUA_OPTIMIZE_STPG		0x01
@@ -677,7 +678,7 @@ static int alua_rtpg(struct scsi_device *sdev, struct alua_port_group *pg)
 	case SCSI_ACCESS_STATE_TRANSITIONING:
 		if (time_before(jiffies, pg->expiry)) {
 			/* State transition, retry */
-			pg->interval = 2;
+			pg->interval = ALUA_RTPG_RETRY_DELAY;
 			err = SCSI_DH_RETRY;
 		} else {
 			struct alua_dh_data *h;
@@ -802,6 +803,8 @@ static void alua_rtpg_work(struct work_struct *work)
 				spin_lock_irqsave(&pg->lock, flags);
 				pg->flags &= ~ALUA_PG_RUNNING;
 				pg->flags |= ALUA_PG_RUN_RTPG;
+				if (!pg->interval)
+					pg->interval = ALUA_RTPG_RETRY_DELAY;
 				spin_unlock_irqrestore(&pg->lock, flags);
 				queue_delayed_work(kaluad_wq, &pg->rtpg_work,
 						   pg->interval * HZ);
@@ -813,6 +816,8 @@ static void alua_rtpg_work(struct work_struct *work)
 		spin_lock_irqsave(&pg->lock, flags);
 		if (err == SCSI_DH_RETRY || pg->flags & ALUA_PG_RUN_RTPG) {
 			pg->flags &= ~ALUA_PG_RUNNING;
+			if (!pg->interval && !(pg->flags & ALUA_PG_RUN_RTPG))
+				pg->interval = ALUA_RTPG_RETRY_DELAY;
 			pg->flags |= ALUA_PG_RUN_RTPG;
 			spin_unlock_irqrestore(&pg->lock, flags);
 			queue_delayed_work(kaluad_wq, &pg->rtpg_work,
Signature · 1 line
2.20.1
author maintainer
#27 lore raw

PATCH 27/30 test_firmware: fix a memory leak bug

In reply to Sasha Levin

From: Wenwen Wang <[email protected]>

[ Upstream commit d4fddac5a51c378c5d3e68658816c37132611e1f ]

In test_firmware_init(), the buffer pointed to by the global pointer 'test_fw_config' is allocated through kzalloc(). Then, the buffer is initialized in __test_firmware_config_init(). In the case that the initialization fails, the following execution in test_firmware_init() needs to be terminated with an error code returned to indicate this failure. However, the allocated buffer is not freed on this execution path, leading to a memory leak bug.

To fix the above issue, free the allocated buffer before returning from test_firmware_init().

---
 lib/test_firmware.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
lib/test_firmware.c +4 -1
index f978aebe60c5b..2e5e18bbfd28e 100644
@@ -895,8 +895,11 @@ static int __init test_firmware_init(void)
 		return -ENOMEM;
 
 	rc = __test_firmware_config_init();
-	if (rc)
+	if (rc) {
+		kfree(test_fw_config);
+		pr_err("could not init firmware test config: %d\n", rc);
 		return rc;
+	}
 
 	rc = misc_register(&test_fw_misc_device);
 	if (rc) {
Signature · 1 line
2.20.1
author maintainer
#28 lore raw

PATCH 28/30 sched/fair: Don't free p->numa_faults with concurrent readers

In reply to Sasha Levin

From: Jann Horn <[email protected]>

[ Upstream commit 16d51a590a8ce3befb1308e0e7ab77f3b661af33 ]

When going through execve(), zero out the NUMA fault statistics instead of freeing them.

During execve, the task is reachable through procfs and the scheduler. A concurrent /proc/*/sched reader can read data from a freed ->numa_faults allocation (confirmed by KASAN) and write it back to userspace. I believe that it would also be possible for a use-after-free read to occur through a race between a NUMA fault and execve(): task_numa_fault() can lead to task_numa_compare(), which invokes task_weight() on the currently running task of a different CPU.

Another way to fix this would be to make ->numa_faults RCU-managed or add extra locking, but it seems easier to wipe the NUMA fault statistics on execve.

Cc Linus Torvalds <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Petr Mladek <[email protected]>
Cc Sergey Senozhatsky <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Will Deacon <[email protected]>
Fixes 82727018b0d3 ("sched/numa: Call task_numa_free() from do_execve()")
---
 fs/exec.c                            |  2 +-
 include/linux/sched/numa_balancing.h |  4 ++--
 kernel/fork.c                        |  2 +-
 kernel/sched/fair.c                  | 24 ++++++++++++++++++++----
 4 files changed, 24 insertions(+), 8 deletions(-)
fs/exec.c +1 -1
index 0936b5a8199ac..4623fc3ac86b8 100644
@@ -1808,7 +1808,7 @@ static int do_execveat_common(int fd, struct filename *filename,
 	current->in_execve = 0;
 	membarrier_execve(current);
 	acct_update_integrals(current);
-	task_numa_free(current);
+	task_numa_free(current, false);
 	free_bprm(bprm);
 	kfree(pathbuf);
 	putname(filename);
include/linux/sched/numa_balancing.h +2 -2
index e7dd04a84ba89..3988762efe15c 100644
@@ -19,7 +19,7 @@
 extern void task_numa_fault(int last_node, int node, int pages, int flags);
 extern pid_t task_numa_group_id(struct task_struct *p);
 extern void set_numabalancing_state(bool enabled);
-extern void task_numa_free(struct task_struct *p);
+extern void task_numa_free(struct task_struct *p, bool final);
 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
 					int src_nid, int dst_cpu);
 #else
@@ -34,7 +34,7 @@ static inline pid_t task_numa_group_id(struct task_struct *p)
 static inline void set_numabalancing_state(bool enabled)
 {
 }
-static inline void task_numa_free(struct task_struct *p)
+static inline void task_numa_free(struct task_struct *p, bool final)
 {
 }
 static inline bool should_numa_migrate_memory(struct task_struct *p,
kernel/fork.c +1 -1
index a5bb8fad54756..919e7cd5cd232 100644
@@ -415,7 +415,7 @@ void __put_task_struct(struct task_struct *tsk)
 	WARN_ON(tsk == current);
 
 	cgroup_free(tsk);
-	task_numa_free(tsk);
+	task_numa_free(tsk, true);
 	security_task_free(tsk);
 	exit_creds(tsk);
 	delayacct_tsk_free(tsk);
kernel/sched/fair.c +20 -4
index af7de1f9906c1..0a4e882d43088 100644
@@ -2358,13 +2358,23 @@ static void task_numa_group(struct task_struct *p, int cpupid, int flags,
 	return;
 }
 
-void task_numa_free(struct task_struct *p)
+/*
+ * Get rid of NUMA staticstics associated with a task (either current or dead).
+ * If @final is set, the task is dead and has reached refcount zero, so we can
+ * safely free all relevant data structures. Otherwise, there might be
+ * concurrent reads from places like load balancing and procfs, and we should
+ * reset the data back to default state without freeing ->numa_faults.
+ */
+void task_numa_free(struct task_struct *p, bool final)
 {
 	struct numa_group *grp = p->numa_group;
-	void *numa_faults = p->numa_faults;
+	unsigned long *numa_faults = p->numa_faults;
 	unsigned long flags;
 	int i;
 
+	if (!numa_faults)
+		return;
+
 	if (grp) {
 		spin_lock_irqsave(&grp->lock, flags);
 		for (i = 0; i < NR_NUMA_HINT_FAULT_STATS * nr_node_ids; i++)
@@ -2377,8 +2387,14 @@ void task_numa_free(struct task_struct *p)
 		put_numa_group(grp);
 	}
 
-	p->numa_faults = NULL;
-	kfree(numa_faults);
+	if (final) {
+		p->numa_faults = NULL;
+		kfree(numa_faults);
+	} else {
+		p->total_numa_faults = 0;
+		for (i = 0; i < NR_NUMA_HINT_FAULT_STATS * nr_node_ids; i++)
+			numa_faults[i] = 0;
+	}
 }
 
 /*
Signature · 1 line
2.20.1
author maintainer
#29 lore raw

PATCH 29/30 tty/ldsem, locking/rwsem: Add missing ACQUIRE to read_failed sleep loop

In reply to Sasha Levin

From: Peter Zijlstra <[email protected]>

[ Upstream commit 952041a8639a7a3a73a2b6573cb8aa8518bc39f8 ]

While reviewing rwsem down_slowpath, Will noticed ldsem had a copy of a bug we just found for rwsem.

  X = 0;

CPU0 CPU1

  rwsem_down_read()
    for (;;) {
      set_current_state(TASK_UNINTERRUPTIBLE);
                        X = 1;
                        rwsem_up_write();
                          rwsem_mark_wake()
                            atomic_long_add(adjustment, &sem->count);
                            smp_store_release(&waiter->task, NULL);

      if (!waiter.task)
        break;

      ...
    }
  r = X;

Allows 'r == 0'.

Cc Linus Torvalds <[email protected]>
Cc Peter Hurley <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Gleixner <[email protected]>
Fixes 4898e640caf0 ("tty: Add timed, writer-prioritized rw semaphore")
---
 drivers/tty/tty_ldsem.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/tty/tty_ldsem.c +2 -3
index 5c2cec298816b..c6ce34161281a 100644
@@ -139,8 +139,7 @@ static void __ldsem_wake_readers(struct ld_semaphore *sem)
 
 	list_for_each_entry_safe(waiter, next, &sem->read_wait, list) {
 		tsk = waiter->task;
-		smp_mb();
-		waiter->task = NULL;
+		smp_store_release(&waiter->task, NULL);
 		wake_up_process(tsk);
 		put_task_struct(tsk);
 	}
@@ -235,7 +234,7 @@ down_read_failed(struct ld_semaphore *sem, long count, long timeout)
 	for (;;) {
 		set_current_state(TASK_UNINTERRUPTIBLE);
 
-		if (!waiter.task)
+		if (!smp_load_acquire(&waiter.task))
 			break;
 		if (!timeout)
 			break;
Signature · 1 line
2.20.1
author maintainer
#30 lore raw

PATCH 30/30 perf/core: Fix creating kernel counters for PMUs that override event->cpu

In reply to Sasha Levin

From: Leonard Crestez <[email protected]>

[ Upstream commit 4ce54af8b33d3e21ca935fc1b89b58cbba956051 ]

Some hardware PMU drivers will override perf_event.cpu inside their event_init callback. This causes a lockdep splat when initialized through the kernel API:

 WARNING: CPU: 0 PID: 250 at kernel/events/core.c:2917 ctx_sched_out+0x78/0x208
 pc : ctx_sched_out+0x78/0x208
 Call trace:
  ctx_sched_out+0x78/0x208
  __perf_install_in_context+0x160/0x248
  remote_function+0x58/0x68
  generic_exec_single+0x100/0x180
  smp_call_function_single+0x174/0x1b8
  perf_install_in_context+0x178/0x188
  perf_event_create_kernel_counter+0x118/0x160

Fix this by calling perf_install_in_context with event->cpu, just like perf_event_open

---
 kernel/events/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/events/core.c +1 -1
index 3d4eb6f840eba..ea4f3f7a0c6f3 100644
@@ -10474,7 +10474,7 @@ perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu,
 		goto err_unlock;
 	}
 
-	perf_install_in_context(ctx, event, cpu);
+	perf_install_in_context(ctx, event, event->cpu);
 	perf_unpin_context(ctx);
 	mutex_unlock(&ctx->mutex);
 
Signature · 1 line
2.20.1