Re: general protection fault in release_urbs

Re: general protection fault in release_urbs

[Takashi Iwai]

On Fri, 30 Sep 2022 04:23:23 +0200,
Sabri N. Ferreiro wrote:
> 
> Hi,
> 
> When I used fuzz testing to test Linux kernel 6.0.0-rc6, the kernel
> triggered the following error:
> HEAD commit: 521a547ced6477c54b4b0cc206000406c221b4d6
> git tree: upstream

Could you retest with 6.0-rc7 or later?
A commit reverting the change might influence on the behavior
significantly.


thanks,

Takashi

> kernel config: https://pastebin.com/raw/hekxU61F
> console log: https://pastebin.com/KVwW9VQs
> 
> It seems that the fuzzer failed to extract any C reproducer, but I
> would so appreciate it if you have any idea how to solve this bug.
> 
> general protection fault, probably for non-canonical address
> 0xdffffc000000000d: 0000 [#1] PREEMPT SMP KASAN
> KASAN: null-ptr-deref in range [0x0000000000000068-0x000000000000006f]
> CPU: 1 PID: 29906 Comm: syz-executor.4 Not tainted 6.0.0-rc6+ #3
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
> 1.13.0-1ubuntu1.1 04/01/2014
> RIP: 0010:release_urb_ctx sound/usb/endpoint.c:97 [inline]
> RIP: 0010:release_urbs sound/usb/endpoint.c:1046 [inline]
> RIP: 0010:release_urbs+0x254/0x5a0 sound/usb/endpoint.c:1031
> Code: 44 89 fe 48 c1 e0 08 4c 8b 74 03 58 e8 75 b4 53 fa 45 85 ff 0f
> 84 29 ff ff ff e8 07 b3 53 fa 49 8d 7e 68 48 89 f8 48 c1 e8 03 <42> 80
> 3c 20 00 0f 85 32 03 00 00 49 8d 7e 60 49 8b 4e 68 48 89 f8
> RSP: 0018:ffffc9001698f8d0 EFLAGS: 00010212
> RAX: 000000000000000d RBX: ffff88805fc44000 RCX: 0000000000040000
> RDX: ffffc900169d1000 RSI: ffff888018c21d40 RDI: 0000000000000068
> RBP: 0000000000000000 R08: ffffffff87273539 R09: 0000000000000000
> R10: 0000000000000005 R11: ffffed100bf88805 R12: dffffc0000000000
> R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000120
> FS: 00007febd6e4e700(0000) GS:ffff88807ec00000(0000) knlGS:0000000000000000
> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 000055555663ddc8 CR3: 0000000065e07000 CR4: 0000000000350ee0
> Call Trace:
> <TASK>
> snd_usb_endpoint_set_params+0x1aab/0x2550
> snd_mask_min include/sound/pcm_params.h:49 [inline]
> params_format include/sound/pcm_params.h:315 [inline]
> snd_usb_hw_params+0x934/0x1180 sound/usb/pcm.c:503
> snd_pcm_hw_params+0xbad/0x1da0 sound/core/pcm_native.c:767
> snd_pcm_kernel_ioctl+0x164/0x310 sound/core/pcm_native.c:3437
> snd_pcm_oss_change_params_locked+0x1834/0x3860 sound/core/oss/pcm_oss.c:976
> snd_pcm_oss_change_params+0x76/0xd0 sound/core/oss/pcm_oss.c:1116
> snd_pcm_oss_make_ready+0xb7/0x170 sound/core/oss/pcm_oss.c:1175
> snd_pcm_oss_get_ptr sound/core/oss/pcm_oss.c:2208 [inline]
> snd_pcm_oss_ioctl+0x3cd/0x3270 sound/core/oss/pcm_oss.c:2729
> vfs_ioctl fs/ioctl.c:51 [inline]
> __do_sys_ioctl fs/ioctl.c:870 [inline]
> __se_sys_ioctl fs/ioctl.c:856 [inline]
> __x64_sys_ioctl+0x193/0x200 fs/ioctl.c:856
> do_syscall_x64 arch/x86/entry/common.c:50 [inline]
> do_syscall_64+0x35/0x80 arch/x86/entry/common.c:80
> entry_SYSCALL_64_after_hwframe+0x63/0xcd
> RIP: 0033:0x7febd66a80fd
> Code: 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 f3 0f 1e fa 48 89 f8 48
> 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d
> 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48
> RSP: 002b:00007febd6e4dbf8 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
> RAX: ffffffffffffffda RBX: 00007febd679c340 RCX: 00007febd66a80fd
> RDX: 00000000200000c0 RSI: 00000000800c5011 RDI: 0000000000000003
> RBP: 00007febd6e4dc50 R08: 0000000000000000 R09: 0000000000000000
> R10: 0000000000000000 R11: 0000000000000246 R12: 000000000000005f
> R13: 00007ffc28c4cf7f R14: 00007ffc28c4d120 R15: 00007febd6e4dd80
> </TASK>
> Modules linked in:
> ---[ end trace 0000000000000000 ]---
> RIP: 0010:release_urb_ctx sound/usb/endpoint.c:97 [inline]
> RIP: 0010:release_urbs sound/usb/endpoint.c:1046 [inline]
> RIP: 0010:release_urbs+0x254/0x5a0 sound/usb/endpoint.c:1031
> Code: 44 89 fe 48 c1 e0 08 4c 8b 74 03 58 e8 75 b4 53 fa 45 85 ff 0f
> 84 29 ff ff ff e8 07 b3 53 fa 49 8d 7e 68 48 89 f8 48 c1 e8 03 <42> 80
> 3c 20 00 0f 85 32 03 00 00 49 8d 7e 60 49 8b 4e 68 48 89 f8
> RSP: 0018:ffffc9001698f8d0 EFLAGS: 00010212
> RAX: 000000000000000d RBX: ffff88805fc44000 RCX: 0000000000040000
> RDX: ffffc900169d1000 RSI: ffff888018c21d40 RDI: 0000000000000068
> RBP: 0000000000000000 R08: ffffffff87273539 R09: 0000000000000000
> R10: 0000000000000005 R11: ffffed100bf88805 R12: dffffc0000000000
> R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000120
> FS: 00007febd6e4e700(0000) GS:ffff88807ec00000(0000) knlGS:0000000000000000
> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 0000001b31424000 CR3: 0000000065e07000 CR4: 0000000000350ee0
> ----------------
> Code disassembly (best guess):
> 0: 44 89 fe mov %r15d,%esi
> 3: 48 c1 e0 08 shl $0x8,%rax
> 7: 4c 8b 74 03 58 mov 0x58(%rbx,%rax,1),%r14
> c: e8 75 b4 53 fa callq 0xfa53b486
> 11: 45 85 ff test %r15d,%r15d
> 14: 0f 84 29 ff ff ff je 0xffffff43
> 1a: e8 07 b3 53 fa callq 0xfa53b326
> 1f: 49 8d 7e 68 lea 0x68(%r14),%rdi
> 23: 48 89 f8 mov %rdi,%rax
> 26: 48 c1 e8 03 shr $0x3,%rax
> * 2a: 42 80 3c 20 00 cmpb $0x0,(%rax,%r12,1) <-- trapping instruction
> 2f: 0f 85 32 03 00 00 jne 0x367
> 35: 49 8d 7e 60 lea 0x60(%r14),%rdi
> 39: 49 8b 4e 68 mov 0x68(%r14),%rcx
> 3d: 48 89 f8 mov %rdi,%rax
> 

Re: general protection fault in release_urbs

[Takashi Iwai]

On Fri, 30 Sep 2022 11:23:21 +0200,
Takashi Iwai wrote:
> 
> On Fri, 30 Sep 2022 04:23:23 +0200,
> Sabri N. Ferreiro wrote:
> > 
> > Hi,
> > 
> > When I used fuzz testing to test Linux kernel 6.0.0-rc6, the kernel
> > triggered the following error:
> > HEAD commit: 521a547ced6477c54b4b0cc206000406c221b4d6
> > git tree: upstream
> 
> Could you retest with 6.0-rc7 or later?
> A commit reverting the change might influence on the behavior
> significantly.

And if it's reproducible on 6.0-rc7, please try the following fix.


thanks,

Takashi

-- 8< --
--- a/sound/usb/endpoint.c
+++ b/sound/usb/endpoint.c
@@ -95,12 +95,13 @@ static inline unsigned get_usb_high_speed_rate(unsigned int rate)
  */
 static void release_urb_ctx(struct snd_urb_ctx *u)
 {
-	if (u->buffer_size)
+	if (u->urb && u->buffer_size)
 		usb_free_coherent(u->ep->chip->dev, u->buffer_size,
 				  u->urb->transfer_buffer,
 				  u->urb->transfer_dma);
 	usb_free_urb(u->urb);
 	u->urb = NULL;
+	u->buffer_size = 0;
 }
 
 static const char *usb_error_string(int err)

Re: general protection fault in release_urbs

[Takashi Iwai]

On Fri, 30 Sep 2022 11:39:03 +0200,
Takashi Iwai wrote:
> 
> On Fri, 30 Sep 2022 11:23:21 +0200,
> Takashi Iwai wrote:
> > 
> > On Fri, 30 Sep 2022 04:23:23 +0200,
> > Sabri N. Ferreiro wrote:
> > > 
> > > Hi,
> > > 
> > > When I used fuzz testing to test Linux kernel 6.0.0-rc6, the kernel
> > > triggered the following error:
> > > HEAD commit: 521a547ced6477c54b4b0cc206000406c221b4d6
> > > git tree: upstream
> > 
> > Could you retest with 6.0-rc7 or later?
> > A commit reverting the change might influence on the behavior
> > significantly.
> 
> And if it's reproducible on 6.0-rc7, please try the following fix.

... and I believe it's happening no matter which version.
I'm going to submit the proper patch, in anyway.


thanks,

Takashi

> 
> 
> thanks,
> 
> Takashi
> 
> -- 8< --
> --- a/sound/usb/endpoint.c
> +++ b/sound/usb/endpoint.c
> @@ -95,12 +95,13 @@ static inline unsigned get_usb_high_speed_rate(unsigned int rate)
>   */
>  static void release_urb_ctx(struct snd_urb_ctx *u)
>  {
> -	if (u->buffer_size)
> +	if (u->urb && u->buffer_size)
>  		usb_free_coherent(u->ep->chip->dev, u->buffer_size,
>  				  u->urb->transfer_buffer,
>  				  u->urb->transfer_dma);
>  	usb_free_urb(u->urb);
>  	u->urb = NULL;
> +	u->buffer_size = 0;
>  }
>  
>  static const char *usb_error_string(int err)