Skip to contentSkip to conversation

intel_th: Fixes for v5.5

GIT PULL Alexander Shishkin 3 participants 8 messages Dec 17, 2019

Subsystems (3)
lore
Share & tools
lore

Alexander Shishkin

author maintainer
#0 lore raw

Hi Greg,

These are the fixes I have so far. Both fixes are for commits in v5.2, so stable is CC'd: one is incorrect irq freeing and one is a NULL dereference from unintended use. Also included 2 new PCI IDs. Signed tag below, individual patches follow. Please consider applying or pullyng. Thanks!

The following changes since commit e42617b825f8073569da76dc4510bfa019b1c35a:

Linux 5.5-rc1 (2019-12-08 14:57:55 -0800)

are available in the Git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/ash/stm.git tags/intel_th-fixes-for-greg-20191216

for you to fetch changes up to 43420a735c271925c74fea8e75c4d6b89fc83bbf:

intel_th: msu: Fix window switching without windows (2019-12-16 19:50:26 +0200)

----------------------------------------------------------------
intel_th: Fixes for v5.5

These are:
* 2 new PCI IDs
* 2 bugfixes for bugs added in v5.2

----------------------------------------------------------------

Alexander Shishkin (4):
intel_th: pci: Add Comet Lake PCH-V support
intel_th: pci: Add Elkhart Lake SOC support
intel_th: Fix freeing IRQs
intel_th: msu: Fix window switching without windows

 drivers/hwtracing/intel_th/core.c     |  7 ++++---
 drivers/hwtracing/intel_th/intel_th.h |  2 ++
 drivers/hwtracing/intel_th/msu.c      | 14 +++++++++-----
 drivers/hwtracing/intel_th/pci.c      | 10 ++++++++++
 4 files changed, 25 insertions(+), 8 deletions(-)
Signature · 1 line
2.24.0
author maintainer
#1 lore raw

[GIT PULL 1/4] intel_th: pci: Add Comet Lake PCH-V support

This adds Intel(R) Trace Hub PCI ID for Comet Lake PCH-V.

---
 drivers/hwtracing/intel_th/pci.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/hwtracing/intel_th/pci.c +5 -0
index ebf3e30e989a..4b2f37578da3 100644
@@ -204,6 +204,11 @@ static const struct pci_device_id intel_th_pci_id_table[] = {
 		PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x06a6),
 		.driver_data = (kernel_ulong_t)&intel_th_2x,
 	},
+	{
+		/* Comet Lake PCH-V */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0xa3a6),
+		.driver_data = (kernel_ulong_t)&intel_th_2x,
+	},
 	{
 		/* Ice Lake NNPI */
 		PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x45c5),
Signature · 1 line
2.24.0
author maintainer
#2 lore raw

[GIT PULL 2/4] intel_th: pci: Add Elkhart Lake SOC support

In reply to Alexander Shishkin

This adds support for Intel Trace Hub in Elkhart Lake.

---
 drivers/hwtracing/intel_th/pci.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/hwtracing/intel_th/pci.c +5 -0
index 4b2f37578da3..e9d90b53bbc4 100644
@@ -234,6 +234,11 @@ static const struct pci_device_id intel_th_pci_id_table[] = {
 		PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x4da6),
 		.driver_data = (kernel_ulong_t)&intel_th_2x,
 	},
+	{
+		/* Elkhart Lake */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x4b26),
+		.driver_data = (kernel_ulong_t)&intel_th_2x,
+	},
 	{ 0 },
 };
 
Signature · 1 line
2.24.0
author maintainer
#3 lore raw

[GIT PULL 3/4] intel_th: Fix freeing IRQs

In reply to Alexander Shishkin

Commit aac8da65174a ("intel_th: msu: Start handling IRQs") implicitly relies on the use of devm_request_irq() to subsequently free the irqs on device removal, but in case of the pci_free_irq_vectors() API, the handlers need to be freed before it is called. Therefore, at the moment the driver's remove path trips a BUG_ON(irq_has_action()):

Quoted 18 lines

kernel BUG at drivers/pci/msi.c:375!
invalid opcode: 0000 1 SMP

CPU: 2 PID: 818 Comm: rmmod Not tainted 5.5.0-rc1+ #1
RIP: 0010:free_msi_irqs+0x67/0x1c0
pci_disable_msi+0x116/0x150
pci_free_irq_vectors+0x1b/0x20
intel_th_pci_remove+0x22/0x30 [intel_th_pci]
pci_device_remove+0x3e/0xb0
device_release_driver_internal+0xf0/0x1c0
driver_detach+0x4c/0x8f
bus_remove_driver+0x5c/0xd0
driver_unregister+0x31/0x50
pci_unregister_driver+0x40/0x90
intel_th_pci_driver_exit+0x10/0xad6 [intel_th_pci]
__x64_sys_delete_module+0x147/0x290
? exit_to_usermode_loop+0xd7/0x120
do_syscall_64+0x57/0x1b0
entry_SYSCALL_64_after_hwframe+0x44/0xa9

Fix this by explicitly freeing irqs before freeing the vectors. We keep using the devm_* variants because they are still useful in early error paths.

Fixes aac8da65174a ("intel_th: msu: Start handling IRQs")
---
 drivers/hwtracing/intel_th/core.c     | 7 ++++---
 drivers/hwtracing/intel_th/intel_th.h | 2 ++
 2 files changed, 6 insertions(+), 3 deletions(-)
drivers/hwtracing/intel_th/core.c +4 -3
index 0dfd97bbde9e..ca232ec565e8 100644
@@ -834,9 +834,6 @@ static irqreturn_t intel_th_irq(int irq, void *data)
 			ret |= d->irq(th->thdev[i]);
 	}
 
-	if (ret == IRQ_NONE)
-		pr_warn_ratelimited("nobody cared for irq\n");
-
 	return ret;
 }
 
@@ -887,6 +884,7 @@ intel_th_alloc(struct device *dev, struct intel_th_drvdata *drvdata,
 
 			if (th->irq == -1)
 				th->irq = devres[r].start;
+			th->num_irqs++;
 			break;
 		default:
 			dev_warn(dev, "Unknown resource type %lx\n",
@@ -940,6 +938,9 @@ void intel_th_free(struct intel_th *th)
 
 	th->num_thdevs = 0;
 
+	for (i = 0; i < th->num_irqs; i++)
+		devm_free_irq(th->dev, th->irq + i, th);
+
 	pm_runtime_get_sync(th->dev);
 	pm_runtime_forbid(th->dev);
 
drivers/hwtracing/intel_th/intel_th.h +2 -0
index 0df480072b6c..6f4f5486fe6d 100644
@@ -261,6 +261,7 @@ enum th_mmio_idx {
  * @num_thdevs:	number of devices in the @thdev array
  * @num_resources:	number of resources in the @resource array
  * @irq:	irq number
+ * @num_irqs:	number of IRQs is use
  * @id:		this Intel TH controller's device ID in the system
  * @major:	device node major for output devices
  */
@@ -277,6 +278,7 @@ struct intel_th {
 	unsigned int		num_thdevs;
 	unsigned int		num_resources;
 	int			irq;
+	int			num_irqs;
 
 	int			id;
 	int			major;
Signature · 1 line
2.24.0
author maintainer
#4 lore raw

[GIT PULL 4/4] intel_th: msu: Fix window switching without windows

In reply to Alexander Shishkin

Commit 6cac7866c2741 ("intel_th: msu: Add a sysfs attribute to trigger window switch") adds a NULL pointer dereference in the case when there are no windows allocated:

Quoted 10 lines
BUG: kernel NULL pointer dereference, address: 0000000000000000
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 0 P4D 0
Oops: 0000 1 SMP
CPU: 5 PID: 1110 Comm: bash Not tainted 5.5.0-rc1+ #1
RIP: 0010:msc_win_switch+0xa/0x80 [intel_th_msu]
Call Trace:
? win_switch_store+0x9b/0xc0 [intel_th_msu]
dev_attr_store+0x17/0x30
Quoted 8 lines
sysfs_kf_write+0x3e/0x50
kernfs_fop_write+0xda/0x1b0
__vfs_write+0x1b/0x40
vfs_write+0xb9/0x1a0
ksys_write+0x67/0xe0
__x64_sys_write+0x1a/0x20
do_syscall_64+0x57/0x1d0
entry_SYSCALL_64_after_hwframe+0x44/0xa9

Fix that by disallowing window switching with multiwindow buffers without windows.

Fixes 6cac7866c2741 ("intel_th: msu: Add a sysfs attribute to trigger window switch")
---
 drivers/hwtracing/intel_th/msu.c | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)
drivers/hwtracing/intel_th/msu.c +9 -5
index 6d240dfae9d9..8e48c7458aa3 100644
@@ -1676,10 +1676,13 @@ static int intel_th_msc_init(struct msc *msc)
 	return 0;
 }
 
-static void msc_win_switch(struct msc *msc)
+static int msc_win_switch(struct msc *msc)
 {
 	struct msc_window *first;
 
+	if (list_empty(&msc->win_list))
+		return -EINVAL;
+
 	first = list_first_entry(&msc->win_list, struct msc_window, entry);
 
 	if (msc_is_last_win(msc->cur_win))
@@ -1691,6 +1694,8 @@ static void msc_win_switch(struct msc *msc)
 	msc->base_addr = msc_win_base_dma(msc->cur_win);
 
 	intel_th_trace_switch(msc->thdev);
+
+	return 0;
 }
 
 /**
@@ -2025,16 +2030,15 @@ win_switch_store(struct device *dev, struct device_attribute *attr,
 	if (val != 1)
 		return -EINVAL;
 
+	ret = -EINVAL;
 	mutex_lock(&msc->buf_mutex);
 	/*
 	 * Window switch can only happen in the "multi" mode.
 	 * If a external buffer is engaged, they have the full
 	 * control over window switching.
 	 */
-	if (msc->mode != MSC_MODE_MULTI || msc->mbuf)
-		ret = -ENOTSUPP;
-	else
-		msc_win_switch(msc);
+	if (msc->mode == MSC_MODE_MULTI && !msc->mbuf)
+		ret = msc_win_switch(msc);
 	mutex_unlock(&msc->buf_mutex);
 
 	return ret ? ret : size;
Signature · 1 line
2.24.0
#5 lore raw

Re: [GIT PULL 4/4] intel_th: msu: Fix window switching without windows

Quoted 3 lines · Alexander Shishkin

Commit 6cac7866c2741 ("intel_th: msu: Add a sysfs attribute to trigger window switch") adds a NULL pointer dereference in the case when there are no windows allocated:

Commit ids should only be specified in 12 digits, not 13 :)

Quoted 20 lines
Quoted 11 lines
BUG: kernel NULL pointer dereference, address: 0000000000000000
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 0 P4D 0
Oops: 0000 1 SMP
CPU: 5 PID: 1110 Comm: bash Not tainted 5.5.0-rc1+ #1
RIP: 0010:msc_win_switch+0xa/0x80 [intel_th_msu]
Call Trace:
? win_switch_store+0x9b/0xc0 [intel_th_msu]
dev_attr_store+0x17/0x30
Quoted 8 lines
sysfs_kf_write+0x3e/0x50
kernfs_fop_write+0xda/0x1b0
__vfs_write+0x1b/0x40
vfs_write+0xb9/0x1a0
ksys_write+0x67/0xe0
__x64_sys_write+0x1a/0x20
do_syscall_64+0x57/0x1d0
entry_SYSCALL_64_after_hwframe+0x44/0xa9
Quoted 5 lines

Fix that by disallowing window switching with multiwindow buffers without windows.

Fixes 6cac7866c2741 ("intel_th: msu: Add a sysfs attribute to trigger window switch")

Same here.

I can go edit it by hand...

#6 lore raw

Re: [GIT PULL 4/4] intel_th: msu: Fix window switching without windows

Hi Greg,

Quoted 7 lines · On Tue, 17 Dec 2019 13:06:29 +0100 Greg Kroah-Hartman…
Quoted 3 lines · Alexander Shishkin

Commit 6cac7866c2741 ("intel_th: msu: Add a sysfs attribute to trigger window switch") adds a NULL pointer dereference in the case when there are no windows allocated:

Commit ids should only be specified in 12 digits, not 13 :)

It is possible that 13 digits may be necessary to be unambiguous within a git repo. In fact, as we continue with a single git repo (Linus'), at some point 13 digits will become normal.

Signature · 2 lines
Cheers,
Stephen Rothwell
author maintainer
#7 lore raw
In reply to Alexander Shishkin
Quoted 6 lines · Alexander Shishkin

Hi Greg,

These are the fixes I have so far. Both fixes are for commits in v5.2, so stable is CC'd: one is incorrect irq freeing and one is a NULL dereference from unintended use. Also included 2 new PCI IDs. Signed tag below, individual patches follow. Please consider applying or pullyng. Thanks!

The tag is wrong, apologies. Should be:

The following changes since commit e42617b825f8073569da76dc4510bfa019b1c35a:

Linux 5.5-rc1 (2019-12-08 14:57:55 -0800)

are available in the Git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/ash/stm.git tags/intel_th-fixes-for-greg-20191216

for you to fetch changes up to 6958619f52e9bc47d7aee0475e9d2c1ae2fcda75:

intel_th: msu: Fix window switching without windows (2019-12-17 13:36:01 +0200)

----------------------------------------------------------------
intel_th: Fixes for v5.5

These are:
* 2 new PCI IDs
* 2 bugfixes for bugs added in v5.2

----------------------------------------------------------------
Alexander Shishkin (4):
intel_th: pci: Add Comet Lake PCH-V support
intel_th: pci: Add Elkhart Lake SOC support
intel_th: Fix freeing IRQs
intel_th: msu: Fix window switching without windows

 drivers/hwtracing/intel_th/core.c     |  7 ++++---
 drivers/hwtracing/intel_th/intel_th.h |  2 ++
 drivers/hwtracing/intel_th/msu.c      | 14 +++++++++-----
 drivers/hwtracing/intel_th/pci.c      | 10 ++++++++++
 4 files changed, 25 insertions(+), 8 deletions(-)