[PATCH] drm/v3d: Free the job and assign it to NULL if initialization fails

Iago Toral itoral at igalia.com
Thu Jan 11 07:13:29 UTC 2024


Ok, thanks for checking, you can add my R-B on the original patch then.

Iago

El mié, 10-01-2024 a las 07:42 -0300, Maira Canal escribió:
> Hi Iago,
> 
> On 1/10/24 03:48, Iago Toral wrote:
> > I think this is fine, but I was wondering if it would be simpler
> > and
> > easier to just remove the sched cleanup from v3d_job_init and
> > instead
> > always rely on callers to eventually call v3d_job_cleanup for fail
> > paths, where we are already calling v3d_job_cleanup.
> 
> If we just remove `drm_sched_job_cleanup()` from `v3d_job_init()` we
> trigger a use-after-free warning by the job refcount:
> 
> [   34.367222] ------------[ cut here ]------------
> [   34.367235] refcount_t: underflow; use-after-free.
> [   34.367274] WARNING: CPU: 0 PID: 1922 at lib/refcount.c:28 
> refcount_warn_saturate+0x108/0x148
> [   34.367298] Modules linked in: rfcomm snd_seq_dummy snd_hrtimer 
> snd_seq snd_seq_device algif_hash aes_neon_bs aes_neon_blk 
> algif_skcipher af_alg bnep hid_logitech_hidpp brcmfmac_wcc brcmfmac 
> hci_uart btbcm bluetooth vc4 brcmutil cfg80211 bcm2835_v4l2(C) 
> bcm2835_mmal_vchiq(C) binfmt_misc snd_soc_hdmi_codec videobuf2_v4l2
> cec 
> ecdh_generic drm_display_helper videobuf2_vmalloc ecc
> videobuf2_memops 
> drm_dma_helper videobuf2_common drm_kms_helper dwc2 raspberrypi_hwmon
> videodev snd_soc_core i2c_brcmstb rfkill libaes hid_logitech_dj mc
> v3d 
> snd_bcm2835(C) i2c_bcm2835 pwm_bcm2835 snd_pcm_dmaengine snd_pcm 
> gpu_sched snd_timer drm_shmem_helper snd nvmem_rmem uio_pdrv_genirq
> uio 
> i2c_dev drm dm_mod fuse drm_panel_orientation_quirks backlight
> configfs 
> ip_tables x_tables ipv6
> [   34.367434] CPU: 0 PID: 1922 Comm: v3d_submit_cl Tainted:
> G         C 
>          6.7.0-rc3-g632ca3c92f38-dirty #74
> [   34.367441] Hardware name: Raspberry Pi 4 Model B Rev 1.5 (DT)
> [   34.367444] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS 
> BTYPE=--)
> [   34.367450] pc : refcount_warn_saturate+0x108/0x148
> [   34.367456] lr : refcount_warn_saturate+0x108/0x148
> [   34.367462] sp : ffffffc08341bb90
> [   34.367465] x29: ffffffc08341bb90 x28: ffffff8102962400 x27: 
> ffffffee5592de88
> [   34.367473] x26: ffffff8116503e00 x25: ffffff81213e8800 x24: 
> 0000000000000048
> [   34.367481] x23: ffffff8101088000 x22: ffffffc08341bcf0 x21: 
> 00000000ffffffea
> [   34.367489] x20: ffffff8102962400 x19: ffffff8102962600 x18: 
> ffffffee5beb3468
> [   34.367497] x17: 0000000000000001 x16: ffffffffffffffff x15: 
> 0000000000000004
> [   34.367504] x14: ffffffee5c163738 x13: 0000000000000fff x12: 
> 0000000000000003
> [   34.367512] x11: 0000000000000000 x10: 0000000000000027 x9 : 
> ada342fc9d5acc00
> [   34.367519] x8 : ada342fc9d5acc00 x7 : 65646e75203a745f x6 : 
> 746e756f63666572
> [   34.367526] x5 : ffffffee5c3129da x4 : ffffffee5c2fc59e x3 : 
> 0000000000000000
> [   34.367533] x2 : 0000000000000000 x1 : ffffffc08341b930 x0 : 
> 0000000000000026
> [   34.367541] Call trace:
> [   34.367544]  refcount_warn_saturate+0x108/0x148
> [   34.367550]  v3d_submit_cl_ioctl+0x4cc/0x5e8 [v3d]
> [   34.367588]  drm_ioctl_kernel+0xe0/0x120 [drm]
> [   34.367767]  drm_ioctl+0x264/0x408 [drm]
> [   34.367866]  __arm64_sys_ioctl+0x9c/0xe0
> [   34.367877]  invoke_syscall+0x4c/0x118
> [   34.367887]  el0_svc_common+0xb8/0xf0
> [   34.367892]  do_el0_svc+0x28/0x40
> [   34.367898]  el0_svc+0x38/0x88
> [   34.367906]  el0t_64_sync_handler+0x84/0x100
> [   34.367913]  el0t_64_sync+0x190/0x198
> [   34.367917] ---[ end trace 0000000000000000 ]---
> 
> Best Regards,
> - Maíra
> 
> > 
> > Iago
> > 
> > El mar, 09-01-2024 a las 11:28 -0300, Maíra Canal escribió:
> > > Currently, if `v3d_job_init()` fails (e.g. in the IGT test "bad-
> > > in-
> > > sync",
> > > where we submit an invalid in-sync to the IOCTL), then we end up
> > > with
> > > the following NULL pointer dereference:
> > > 
> > > [   34.146279] Unable to handle kernel NULL pointer dereference
> > > at
> > > virtual address 0000000000000078
> > > [   34.146301] Mem abort info:
> > > [   34.146306]   ESR = 0x0000000096000005
> > > [   34.146315]   EC = 0x25: DABT (current EL), IL = 32 bits
> > > [   34.146322]   SET = 0, FnV = 0
> > > [   34.146328]   EA = 0, S1PTW = 0
> > > [   34.146334]   FSC = 0x05: level 1 translation fault
> > > [   34.146340] Data abort info:
> > > [   34.146345]   ISV = 0, ISS = 0x00000005, ISS2 = 0x00000000
> > > [   34.146351]   CM = 0, WnR = 0, TnD = 0, TagAccess = 0
> > > [   34.146357]   GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
> > > [   34.146366] user pgtable: 4k pages, 39-bit VAs,
> > > pgdp=00000001232e6000
> > > [   34.146375] [0000000000000078] pgd=0000000000000000,
> > > p4d=0000000000000000, pud=0000000000000000
> > > [   34.146399] Internal error: Oops: 0000000096000005 [#1]
> > > PREEMPT
> > > SMP
> > > [   34.146406] Modules linked in: rfcomm snd_seq_dummy
> > > snd_hrtimer
> > > snd_seq snd_seq_device algif_hash aes_neon_bs aes_neon_blk
> > > algif_skcipher af_alg bnep hid_logitech_hidpp brcmfmac_wcc
> > > brcmfmac
> > > brcmutil hci_uart vc4 btbcm cfg80211 bluetooth bcm2835_v4l2(C)
> > > snd_soc_hdmi_codec binfmt_misc cec drm_display_helper
> > > hid_logitech_dj
> > > bcm2835_mmal_vchiq(C) drm_dma_helper drm_kms_helper
> > > videobuf2_v4l2
> > > raspberrypi_hwmon ecdh_generic videobuf2_vmalloc videobuf2_memops
> > > ecc
> > > videobuf2_common rfkill videodev libaes snd_soc_core dwc2
> > > i2c_brcmstb
> > > snd_pcm_dmaengine snd_bcm2835(C) i2c_bcm2835 pwm_bcm2835 snd_pcm
> > > mc
> > > v3d snd_timer snd gpu_sched drm_shmem_helper nvmem_rmem
> > > uio_pdrv_genirq uio i2c_dev drm fuse dm_mod
> > > drm_panel_orientation_quirks backlight configfs ip_tables
> > > x_tables
> > > ipv6
> > > [   34.146556] CPU: 1 PID: 1890 Comm: v3d_submit_csd Tainted:
> > > G         C         6.7.0-rc3-g49ddab089611 #68
> > > [   34.146563] Hardware name: Raspberry Pi 4 Model B Rev 1.5 (DT)
> > > [   34.146569] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -
> > > SSBS
> > > BTYPE=--)
> > > [   34.146575] pc : drm_sched_job_cleanup+0x3c/0x190 [gpu_sched]
> > > [   34.146611] lr : v3d_submit_csd_ioctl+0x1b4/0x460 [v3d]
> > > [   34.146653] sp : ffffffc083cbbb80
> > > [   34.146658] x29: ffffffc083cbbb90 x28: ffffff81035afc00 x27:
> > > ffffffe77a641168
> > > [   34.146668] x26: ffffff81056a8000 x25: 0000000000000058 x24:
> > > 0000000000000000
> > > [   34.146677] x23: ffffff81065e2000 x22: ffffff81035afe00 x21:
> > > ffffffc083cbbcf0
> > > [   34.146686] x20: ffffff81035afe00 x19: 00000000ffffffea x18:
> > > 0000000000000000
> > > [   34.146694] x17: 0000000000000000 x16: ffffffe7989e34b0 x15:
> > > 0000000000000000
> > > [   34.146703] x14: 0000000004000004 x13: ffffff81035afe80 x12:
> > > ffffffc083cb8000
> > > [   34.146711] x11: cc57e05dfbe5ef00 x10: cc57e05dfbe5ef00 x9 :
> > > ffffffe77a64131c
> > > [   34.146719] x8 : 0000000000000000 x7 : 0000000000000000 x6 :
> > > 000000000000003f
> > > [   34.146727] x5 : 0000000000000040 x4 : ffffff81fefb03f0 x3 :
> > > ffffffc083cbba40
> > > [   34.146736] x2 : ffffff81056a8000 x1 : ffffffe7989e35e8 x0 :
> > > 0000000000000000
> > > [   34.146745] Call trace:
> > > [   34.146748]  drm_sched_job_cleanup+0x3c/0x190 [gpu_sched]
> > > [   34.146768]  v3d_submit_csd_ioctl+0x1b4/0x460 [v3d]
> > > [   34.146791]  drm_ioctl_kernel+0xe0/0x120 [drm]
> > > [   34.147029]  drm_ioctl+0x264/0x408 [drm]
> > > [   34.147135]  __arm64_sys_ioctl+0x9c/0xe0
> > > [   34.147152]  invoke_syscall+0x4c/0x118
> > > [   34.147162]  el0_svc_common+0xb8/0xf0
> > > [   34.147168]  do_el0_svc+0x28/0x40
> > > [   34.147174]  el0_svc+0x38/0x88
> > > [   34.147184]  el0t_64_sync_handler+0x84/0x100
> > > [   34.147191]  el0t_64_sync+0x190/0x198
> > > [   34.147201] Code: aa0003f4 f90007e8 f9401008 aa0803e0
> > > (b8478c09)
> > > [   34.147210] ---[ end trace 0000000000000000 ]---
> > > 
> > > This happens because we are calling `drm_sched_job_cleanup()`
> > > twice:
> > > once at `v3d_job_init()` and again when we call
> > > `v3d_job_cleanup()`.
> > > 
> > > To mitigate this issue, we can return to the same approach that
> > > we
> > > used
> > > to use before 464c61e76de8: deallocate the job after
> > > `v3d_job_init()`
> > > fails and assign it to NULL. Then, when we call
> > > `v3d_job_cleanup()`,
> > > job
> > > is NULL and the function returns.
> > > 
> > > Fixes: 464c61e76de8 ("drm/v3d: Decouple job allocation from job
> > > initiation")
> > > Signed-off-by: Maíra Canal <mcanal at igalia.com>
> > > ---
> > >   drivers/gpu/drm/v3d/v3d_submit.c | 35
> > > +++++++++++++++++++++++++-----
> > > --
> > >   1 file changed, 28 insertions(+), 7 deletions(-)
> > > 
> > > diff --git a/drivers/gpu/drm/v3d/v3d_submit.c
> > > b/drivers/gpu/drm/v3d/v3d_submit.c
> > > index fcff41dd2315..88f63d526b22 100644
> > > --- a/drivers/gpu/drm/v3d/v3d_submit.c
> > > +++ b/drivers/gpu/drm/v3d/v3d_submit.c
> > > @@ -147,6 +147,13 @@ v3d_job_allocate(void **container, size_t
> > > size)
> > >   	return 0;
> > >   }
> > > 
> > > +static void
> > > +v3d_job_deallocate(void **container)
> > > +{
> > > +	kfree(*container);
> > > +	*container = NULL;
> > > +}
> > > +
> > >   static int
> > >   v3d_job_init(struct v3d_dev *v3d, struct drm_file *file_priv,
> > >   	     struct v3d_job *job, void (*free)(struct kref
> > > *ref),
> > > @@ -273,8 +280,10 @@ v3d_setup_csd_jobs_and_bos(struct drm_file
> > > *file_priv,
> > > 
> > >   	ret = v3d_job_init(v3d, file_priv, &(*job)->base,
> > >   			   v3d_job_free, args->in_sync, se,
> > > V3D_CSD);
> > > -	if (ret)
> > > +	if (ret) {
> > > +		v3d_job_deallocate((void *)job);
> > >   		return ret;
> > > +	}
> > > 
> > >   	ret = v3d_job_allocate((void *)clean_job,
> > > sizeof(**clean_job));
> > >   	if (ret)
> > > @@ -282,8 +291,10 @@ v3d_setup_csd_jobs_and_bos(struct drm_file
> > > *file_priv,
> > > 
> > >   	ret = v3d_job_init(v3d, file_priv, *clean_job,
> > >   			   v3d_job_free, 0, NULL,
> > > V3D_CACHE_CLEAN);
> > > -	if (ret)
> > > +	if (ret) {
> > > +		v3d_job_deallocate((void *)clean_job);
> > >   		return ret;
> > > +	}
> > > 
> > >   	(*job)->args = *args;
> > > 
> > > @@ -860,8 +871,10 @@ v3d_submit_cl_ioctl(struct drm_device *dev,
> > > void
> > > *data,
> > > 
> > >   	ret = v3d_job_init(v3d, file_priv, &render->base,
> > >   			   v3d_render_job_free, args-
> > > >in_sync_rcl,
> > > &se, V3D_RENDER);
> > > -	if (ret)
> > > +	if (ret) {
> > > +		v3d_job_deallocate((void *)&render);
> > >   		goto fail;
> > > +	}
> > > 
> > >   	render->start = args->rcl_start;
> > >   	render->end = args->rcl_end;
> > > @@ -874,8 +887,10 @@ v3d_submit_cl_ioctl(struct drm_device *dev,
> > > void
> > > *data,
> > > 
> > >   		ret = v3d_job_init(v3d, file_priv, &bin->base,
> > >   				   v3d_job_free, args-
> > > >in_sync_bcl,
> > > &se, V3D_BIN);
> > > -		if (ret)
> > > +		if (ret) {
> > > +			v3d_job_deallocate((void *)&bin);
> > >   			goto fail;
> > > +		}
> > > 
> > >   		bin->start = args->bcl_start;
> > >   		bin->end = args->bcl_end;
> > > @@ -892,8 +907,10 @@ v3d_submit_cl_ioctl(struct drm_device *dev,
> > > void
> > > *data,
> > > 
> > >   		ret = v3d_job_init(v3d, file_priv, clean_job,
> > >   				   v3d_job_free, 0, NULL,
> > > V3D_CACHE_CLEAN);
> > > -		if (ret)
> > > +		if (ret) {
> > > +			v3d_job_deallocate((void *)&clean_job);
> > >   			goto fail;
> > > +		}
> > > 
> > >   		last_job = clean_job;
> > >   	} else {
> > > @@ -1015,8 +1032,10 @@ v3d_submit_tfu_ioctl(struct drm_device
> > > *dev,
> > > void *data,
> > > 
> > >   	ret = v3d_job_init(v3d, file_priv, &job->base,
> > >   			   v3d_job_free, args->in_sync, &se,
> > > V3D_TFU);
> > > -	if (ret)
> > > +	if (ret) {
> > > +		v3d_job_deallocate((void *)&job);
> > >   		goto fail;
> > > +	}
> > > 
> > >   	job->base.bo = kcalloc(ARRAY_SIZE(args->bo_handles),
> > >   			       sizeof(*job->base.bo),
> > > GFP_KERNEL);
> > > @@ -1233,8 +1252,10 @@ v3d_submit_cpu_ioctl(struct drm_device
> > > *dev,
> > > void *data,
> > > 
> > >   	ret = v3d_job_init(v3d, file_priv, &cpu_job->base,
> > >   			   v3d_job_free, 0, &se, V3D_CPU);
> > > -	if (ret)
> > > +	if (ret) {
> > > +		v3d_job_deallocate((void *)&cpu_job);
> > >   		goto fail;
> > > +	}
> > > 
> > >   	clean_job = cpu_job->indirect_csd.clean_job;
> > >   	csd_job = cpu_job->indirect_csd.job;
> > > --
> > > 2.43.0
> > > 
> > > 
> > 
> 



More information about the dri-devel mailing list