diff mbox series

[RESEND] drm: fix crash in drm_minor_alloc_release

Message ID 20221107144500.3692212-1-stanislaw.gruszka@linux.intel.com (mailing list archive)
State New, archived
Headers show
Series [RESEND] drm: fix crash in drm_minor_alloc_release | expand

Commit Message

Stanislaw Gruszka Nov. 7, 2022, 2:45 p.m. UTC 
If drm_sysfs_minor_alloc() fail in drm_minor_alloc() we can end up
freeing invalid minor->kdev pointer and drm_minor_alloc_release()
will crash like below:

RIP: 0010:kobject_put+0x19/0x1c0
RSP: 0018:ffffbc7001637c38 EFLAGS: 00010282
RAX: ffffffffa8d6deb0 RBX: 00000000ffffffff RCX: ffff9cb5912d4540
RDX: ffffffffa9c45ec5 RSI: ffff9cb5902f2b68 RDI: fffffffffffffff4
RBP: fffffffffffffff4 R08: ffffffffa9c40dec R09: 0000000000000008
R10: ffffffffaa81f7d2 R11: 00000000aa81f7ca R12: ffff9cb5912d4540
R13: ffff9cb5912d4540 R14: dead000000000122 R15: dead000000000100
FS:  00007f56b06e6740(0000) GS:ffff9cb728b40000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000030 CR3: 000000011285b004 CR4: 0000000000170ee0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff07f0 DR7: 0000000000000400
Call Trace:
 <TASK>
 drm_minor_alloc_release+0x19/0x50
 drm_managed_release+0xab/0x150
 drm_dev_init+0x21f/0x2f0
 __devm_drm_dev_alloc+0x3c/0xa0
 ivpu_probe+0x59/0x797 [intel_vpu 127058409b05eb2f99dcdecd3330bee28d6b3e76]
 pci_device_probe+0xa4/0x160
 really_probe+0x164/0x340
 __driver_probe_device+0x10d/0x190
 device_driver_attach+0x26/0x50
 bind_store+0x9f/0x120
 kernfs_fop_write_iter+0x12d/0x1c0
 new_sync_write+0x106/0x180
 vfs_write+0x216/0x2a0
 ksys_write+0x65/0xe0
 do_syscall_64+0x35/0x80
 entry_SYSCALL_64_after_hwframe+0x44/0xae

Fix this crash by checking minor->kdev when freeing.

Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
---
 drivers/gpu/drm/drm_drv.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

Comments

Ville Syrjälä Nov. 7, 2022, 3:10 p.m. UTC | #1 
On Mon, Nov 07, 2022 at 03:45:00PM +0100, Stanislaw Gruszka wrote:
> If drm_sysfs_minor_alloc() fail in drm_minor_alloc() we can end up
> freeing invalid minor->kdev pointer and drm_minor_alloc_release()
> will crash like below:
> 
> RIP: 0010:kobject_put+0x19/0x1c0
> RSP: 0018:ffffbc7001637c38 EFLAGS: 00010282
> RAX: ffffffffa8d6deb0 RBX: 00000000ffffffff RCX: ffff9cb5912d4540
> RDX: ffffffffa9c45ec5 RSI: ffff9cb5902f2b68 RDI: fffffffffffffff4
> RBP: fffffffffffffff4 R08: ffffffffa9c40dec R09: 0000000000000008
> R10: ffffffffaa81f7d2 R11: 00000000aa81f7ca R12: ffff9cb5912d4540
> R13: ffff9cb5912d4540 R14: dead000000000122 R15: dead000000000100
> FS:  00007f56b06e6740(0000) GS:ffff9cb728b40000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 0000000000000030 CR3: 000000011285b004 CR4: 0000000000170ee0
> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> DR3: 0000000000000000 DR6: 00000000ffff07f0 DR7: 0000000000000400
> Call Trace:
>  <TASK>
>  drm_minor_alloc_release+0x19/0x50
>  drm_managed_release+0xab/0x150
>  drm_dev_init+0x21f/0x2f0
>  __devm_drm_dev_alloc+0x3c/0xa0
>  ivpu_probe+0x59/0x797 [intel_vpu 127058409b05eb2f99dcdecd3330bee28d6b3e76]
>  pci_device_probe+0xa4/0x160
>  really_probe+0x164/0x340
>  __driver_probe_device+0x10d/0x190
>  device_driver_attach+0x26/0x50
>  bind_store+0x9f/0x120
>  kernfs_fop_write_iter+0x12d/0x1c0
>  new_sync_write+0x106/0x180
>  vfs_write+0x216/0x2a0
>  ksys_write+0x65/0xe0
>  do_syscall_64+0x35/0x80
>  entry_SYSCALL_64_after_hwframe+0x44/0xae
> 
> Fix this crash by checking minor->kdev when freeing.
> 
> Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
> ---
>  drivers/gpu/drm/drm_drv.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/drm_drv.c b/drivers/gpu/drm/drm_drv.c
> index 8214a0b1ab7f..e3a1243dd2ae 100644
> --- a/drivers/gpu/drm/drm_drv.c
> +++ b/drivers/gpu/drm/drm_drv.c
> @@ -102,7 +102,8 @@ static void drm_minor_alloc_release(struct drm_device *dev, void *data)
>  
>  	WARN_ON(dev != minor->dev);
>  
> -	put_device(minor->kdev);
> +	if (!IS_ERR(minor->kdev))
> +		put_device(minor->kdev);

Assigning error pointers into things is a terrible idea.
IMO the correct fix would be to not return some
half-constructed garbage from drm_minor_alloc().
So basically should at least partically revert
commit f96306f9892b ("drm: manage drm_minor cleanup with drmm_")

>  
>  	spin_lock_irqsave(&drm_minor_lock, flags);
>  	idr_remove(&drm_minors_idr, minor->index);
> -- 
> 2.25.1
Stanislaw Gruszka Nov. 7, 2022, 3:40 p.m. UTC | #2 
On Mon, Nov 07, 2022 at 05:10:48PM +0200, Ville Syrjälä wrote:
> On Mon, Nov 07, 2022 at 03:45:00PM +0100, Stanislaw Gruszka wrote:
> > index 8214a0b1ab7f..e3a1243dd2ae 100644
> > --- a/drivers/gpu/drm/drm_drv.c
> > +++ b/drivers/gpu/drm/drm_drv.c
> > @@ -102,7 +102,8 @@ static void drm_minor_alloc_release(struct drm_device *dev, void *data)
> >  
> >  	WARN_ON(dev != minor->dev);
> >  
> > -	put_device(minor->kdev);
> > +	if (!IS_ERR(minor->kdev))
> > +		put_device(minor->kdev);
> 
> Assigning error pointers into things is a terrible idea.
> IMO the correct fix would be to not return some
> half-constructed garbage from drm_minor_alloc().
> So basically should at least partically revert
> commit f96306f9892b ("drm: manage drm_minor cleanup with drmm_")

I would prefer to not change any ordering or remove drmm_* stuff, since
as pointed to above commit message, things are tricky there.

I think assigning NULL to minor->kdev should be fine:

	if (IS_ERR(minor->kdev)) {
		r = PTR_ERR(minor->kdev);
		minor->kdev = NULL;
		return r;
	}

put_device() in drm_minor_alloc_release() will cope nicely with it.

Regards
Stanislaw
Ville Syrjälä Nov. 7, 2022, 3:56 p.m. UTC | #3 
On Mon, Nov 07, 2022 at 04:40:41PM +0100, Stanislaw Gruszka wrote:
> On Mon, Nov 07, 2022 at 05:10:48PM +0200, Ville Syrjälä wrote:
> > On Mon, Nov 07, 2022 at 03:45:00PM +0100, Stanislaw Gruszka wrote:
> > > index 8214a0b1ab7f..e3a1243dd2ae 100644
> > > --- a/drivers/gpu/drm/drm_drv.c
> > > +++ b/drivers/gpu/drm/drm_drv.c
> > > @@ -102,7 +102,8 @@ static void drm_minor_alloc_release(struct drm_device *dev, void *data)
> > >  
> > >  	WARN_ON(dev != minor->dev);
> > >  
> > > -	put_device(minor->kdev);
> > > +	if (!IS_ERR(minor->kdev))
> > > +		put_device(minor->kdev);
> > 
> > Assigning error pointers into things is a terrible idea.
> > IMO the correct fix would be to not return some
> > half-constructed garbage from drm_minor_alloc().
> > So basically should at least partically revert
> > commit f96306f9892b ("drm: manage drm_minor cleanup with drmm_")
> 
> I would prefer to not change any ordering or remove drmm_* stuff, since
> as pointed to above commit message, things are tricky there.

Looks to me that it's only tricky because of drmm. Without that it was
totally clear what was happening. I think if the managed stuff is making
stuff more tricky then it has failed its purpose.

> 
> I think assigning NULL to minor->kdev should be fine:
> 
> 	if (IS_ERR(minor->kdev)) {
> 		r = PTR_ERR(minor->kdev);
> 		minor->kdev = NULL;
> 		return r;
> 	}
> 
> put_device() in drm_minor_alloc_release() will cope nicely with it.
> 
> Regards
> Stanislaw
Stanislaw Gruszka Nov. 8, 2022, 4:24 p.m. UTC | #4 
On Mon, Nov 07, 2022 at 05:56:36PM +0200, Ville Syrjälä wrote:
> On Mon, Nov 07, 2022 at 04:40:41PM +0100, Stanislaw Gruszka wrote:
> > On Mon, Nov 07, 2022 at 05:10:48PM +0200, Ville Syrjälä wrote:
> > > On Mon, Nov 07, 2022 at 03:45:00PM +0100, Stanislaw Gruszka wrote:
> > > > index 8214a0b1ab7f..e3a1243dd2ae 100644
> > > > --- a/drivers/gpu/drm/drm_drv.c
> > > > +++ b/drivers/gpu/drm/drm_drv.c
> > > > @@ -102,7 +102,8 @@ static void drm_minor_alloc_release(struct drm_device *dev, void *data)
> > > >  
> > > >  	WARN_ON(dev != minor->dev);
> > > >  
> > > > -	put_device(minor->kdev);
> > > > +	if (!IS_ERR(minor->kdev))
> > > > +		put_device(minor->kdev);
> > > 
> > > Assigning error pointers into things is a terrible idea.
> > > IMO the correct fix would be to not return some
> > > half-constructed garbage from drm_minor_alloc().
> > > So basically should at least partically revert
> > > commit f96306f9892b ("drm: manage drm_minor cleanup with drmm_")
> > 
> > I would prefer to not change any ordering or remove drmm_* stuff, since
> > as pointed to above commit message, things are tricky there.
> 
> Looks to me that it's only tricky because of drmm. Without that it was
> totally clear what was happening. I think if the managed stuff is making
> stuff more tricky then it has failed its purpose.

I'm not huge fan of managed resources everywhere either, but I think
we should do rather small fixes for bugs to avoid regressions.

> > I think assigning NULL to minor->kdev should be fine:
> > 
> > 	if (IS_ERR(minor->kdev)) {
> > 		r = PTR_ERR(minor->kdev);
> > 		minor->kdev = NULL;
> > 		return r;
> > 	}

Seems having minor->kdev NULL was ordinal Daniel idea in commit
f96306f9892b, but was not done properly when finally patch get's in.

Regards
Stanislaw
diff mbox series

Patch

diff --git a/drivers/gpu/drm/drm_drv.c b/drivers/gpu/drm/drm_drv.c
index 8214a0b1ab7f..e3a1243dd2ae 100644
--- a/drivers/gpu/drm/drm_drv.c
+++ b/drivers/gpu/drm/drm_drv.c
@@ -102,7 +102,8 @@  static void drm_minor_alloc_release(struct drm_device *dev, void *data)
 
 	WARN_ON(dev != minor->dev);
 
-	put_device(minor->kdev);
+	if (!IS_ERR(minor->kdev))
+		put_device(minor->kdev);
 
 	spin_lock_irqsave(&drm_minor_lock, flags);
 	idr_remove(&drm_minors_idr, minor->index);