Bug 71081

Summary:	kernel: kernel BUG at drivers/gpu/drm/radeon/radeon_sa.c:322!
Product:	Drivers	Reporter:	William Shuman (wshuman3)
Component:	Video(DRI - non Intel)	Assignee:	drivers_video-dri
Status:	NEW ---
Severity:	normal	CC:	alan, alexdeucher, deathsimple, szg00000
Priority:	P1
Hardware:	x86-64
OS:	Linux
Kernel Version:	3.13.5	Subsystem:
Regression:	No	Bisected commit-id:
Attachments:	journal journal_wip

Description William Shuman 2014-02-24 13:24:43 UTC

: [drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation -12!
: [drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation -12!
: ------------[ cut here ]------------
: kernel BUG at drivers/gpu/drm/radeon/radeon_sa.c:322!
: invalid opcode: 0000 [#1] PREEMPT SMP 
: Modules linked in: tun rfcomm fuse ipt_MASQUERADE iptable_nat nf_nat_ipv4 nf_nat nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack ipt_REJECT xt_CHECKSUM iptable_
:  pps_core parport_pc hp_accel kvm_intel mei_me lis3lv02d wmi mei input_polldev soundcore parport thermal battery lpc_ich video ac button kvm processor microcode ext4 crc16
: CPU: 0 PID: 583 Comm: Xorg Not tainted 3.13.5-1-ARCH #1
: Hardware name: Hewlett-Packard HP EliteBook 8570w/176B, BIOS 68IAV Ver. F.43 10/08/2013
: task: ffff8804255b7500 ti: ffff88041e904000 task.ti: ffff88041e904000
: RIP: 0010:[<ffffffffa0adee5d>]  [<ffffffffa0adee5d>] radeon_sa_bo_new+0x49d/0x4c0 [radeon]
: RSP: 0018:ffff88041e9059e8  EFLAGS: 00010287
: RAX: 0000000000102d10 RBX: ffff880426ac1360 RCX: 0000000000102d10
: RDX: ffff88041e905b58 RSI: ffff880426ac1360 RDI: ffff880426ac0000
: RBP: ffff88041e905ac8 R08: 0000000000000100 R09: 0000000000000001
: R10: 0000000000000000 R11: ffffffffa0ade3b5 R12: ffff880426ac0000
: R13: 0000000000000000 R14: 0000000000000003 R15: ffff8800b9384380
: FS:  00007fc8e0184700(0000) GS:ffff88043dc00000(0000) knlGS:0000000000000000
: CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
: CR2: 00007f906699f840 CR3: 0000000425f19000 CR4: 00000000001407e0
: Stack:
:  ffff8804255b7500 ffff880400000008 ffff880426ac0000 ffff88041e905b58
:  00102d1001000000 0000000000000000 0000000000000100 0000000000000246
:  ffff880426ac0f68 0000000000000000 ffff8800701cac40 000000000030f80f
: Call Trace:
:  [<ffffffffa0a944e9>] radeon_ib_get+0x39/0xe0 [radeon]
:  [<ffffffffa0a82401>] radeon_vm_bo_update+0x101/0x6a0 [radeon]
:  [<ffffffffa09f8dcd>] ? ttm_eu_list_ref_sub+0x3d/0x60 [ttm]
:  [<ffffffffa0a96f6a>] radeon_cs_ioctl+0x90a/0xa20 [radeon]
:  [<ffffffffa08e6bf2>] drm_ioctl+0x502/0x630 [drm]
:  [<ffffffffa0a6304e>] radeon_drm_ioctl+0x4e/0x90 [radeon]
:  [<ffffffff811b5fe0>] do_vfs_ioctl+0x2e0/0x4c0
:  [<ffffffff811b6241>] SyS_ioctl+0x81/0xa0
:  [<ffffffff8152142d>] system_call_fastpath+0x1a/0x1f
: Code: e0 48 8b 9d 38 ff ff ff 48 8b 3b e8 1e ab 6a e0 48 c7 03 00 00 00 00 44 89 e8 e9 0d fd ff ff b8 f4 ff ff ff e9 03 fd ff ff 0f 0b <0f> 0b 4c 8b a5 58 ff ff ff 41 bd 0
: RIP  [<ffffffffa0adee5d>] radeon_sa_bo_new+0x49d/0x4c0 [radeon]
:  RSP <ffff88041e9059e8>
: ---[ end trace ffdd33bb64df5e39 ]---

Comment 1 Alex Deucher 2014-02-24 22:46:53 UTC

Please attach your full dmesg output.  Was there a particular app or operation that triggered this?

Comment 2 William Shuman 2014-02-25 02:38:54 UTC

Created attachment 127351 [details]
journal

Attaching. 

This happened while using firefox and browsing google maps in street view.

Comment 3 Christian König 2014-02-25 09:46:42 UTC

That's a known problem with the current VM implementation. The application just tries to use a buffer larger than what we can handle in the VM code, but still small enough to not trigger the error path.

Please try the drm-next-3.15-wip branch from git://people.freedesktop.org/~deathsimple/linux and see if you can still reproduce the problem.

Thanks in advance,
Christian.

Comment 4 William Shuman 2014-02-25 16:28:42 UTC

Created attachment 127451 [details]
journal_wip

Attaching journal from drm-next-3.15-wip

Comment 5 Christian König 2014-02-25 16:36:26 UTC

(In reply to William Shuman from comment #4)
> Created attachment 127451 [details]
> journal_wip
> 
> Attaching journal from drm-next-3.15-wip

Thanks, that was helpful.

The initial bug is gone, but we now see new bugs with VM handling. Probably need to work a bit more on that patchset.