RE: [PATCH] drm/amdgpu: Fix Kernel Oops triggered by kfdtest

[Kuehling, Felix]

Apologies. We already have a fix for this on our internal amd-kfd-staging branch, but it's missing from amd-staging-drm-next. I'll cherry-pick our fix to amd-staging-drm-next and nominate it for drm-fixes.

Regards,
  Felix

-----Original Message-----
From: amd-gfx <amd-gfx-bounces@xxxxxxxxxxxxxxxxxxxxx> On Behalf Of Joerg Roedel
Sent: Thursday, November 15, 2018 9:46 AM
To: Zhu, Rex <Rex.Zhu@xxxxxxx>; Quan, Evan <Evan.Quan@xxxxxxx>; Deucher, Alexander <Alexander.Deucher@xxxxxxx>; Koenig, Christian <Christian.Koenig@xxxxxxx>; Zhou, David(ChunMing) <David1.Zhou@xxxxxxx>
Cc: Joerg Roedel <jroedel@xxxxxxx>; dri-devel@xxxxxxxxxxxxxxxxxxxxx; amd-gfx@xxxxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx
Subject: [PATCH] drm/amdgpu: Fix Kernel Oops triggered by kfdtest

From: Joerg Roedel <jroedel@xxxxxxx>

Running kfdtest on Kaveri triggers a kernel NULL-ptr dereference:

	BUG: unable to handle kernel NULL pointer dereference at 0000000000000000
	PGD 42c017067 P4D 42c017067 PUD 40f071067 PMD 0
	Oops: 0010 [#1] SMP NOPTI
	CPU: 0 PID: 13107 Comm: kfdtest Not tainted 4.20.0-rc2+ #11
	Hardware name: Gigabyte Technology Co., Ltd. To be filled by O.E.M./F2A88XM-HD3, BIOS F6 05/28/2014
	RIP: 0010:          (null)
	Code: Bad RIP value.
	RSP: 0018:ffffc90001adbbf0 EFLAGS: 00010202
	RAX: ffffffffa0806240 RBX: ffff88842a0fbc00 RCX: 0000000000000002
	RDX: 0000000000000001 RSI: 0000000000000004 RDI: ffff888429690000
	RBP: ffffc90001adbbf8 R08: 0000000000002000 R09: ffff88842e542ec0
	R10: 00007feff778f008 R11: 00007feff778f010 R12: 0000000000000000
	R13: ffff88840f063a20 R14: ffff88842a0fbd20 R15: 000000000f43ff60
	FS:  00007feff7769740(0000) GS:ffff88842fa00000(0000) knlGS:0000000000000000
	CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
	CR2: ffffffffffffffd6 CR3: 000000040f122000 CR4: 00000000000406f0
	Call Trace:
	 ? amdgpu_amdkfd_set_compute_idle+0x29/0x30 [amdgpu]
	 register_process+0x140/0x150 [amdgpu]
	 pqm_create_queue+0x395/0x560 [amdgpu]
	 kfd_ioctl_create_queue+0x285/0x680 [amdgpu]
	 kfd_ioctl+0x27f/0x450 [amdgpu]
	 ? kfd_ioctl_destroy_queue+0x80/0x80 [amdgpu]
	 do_vfs_ioctl+0x96/0x6a0
	 ? __audit_syscall_entry+0xdd/0x130
	 ? handle_mm_fault+0x11b/0x240
	 ksys_ioctl+0x67/0x90
	 __x64_sys_ioctl+0x1a/0x20
	 do_syscall_64+0x61/0x190
	 entry_SYSCALL_64_after_hwframe+0x44/0xa9

The reason is that the pp_funcs->switch_power_profile function pointer is not set for a Kaveri ASIC and thus the kernel calls a NULL pointer.

Add a check before calling the function to avoid that.

Signed-off-by: Joerg Roedel <jroedel@xxxxxxx>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
index f972cd156795..0ecedd30f2aa 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
@@ -337,8 +337,9 @@ enum amdgpu_pcie_gen {
 			(adev)->powerplay.pp_handle, request))
 
 #define amdgpu_dpm_switch_power_profile(adev, type, en) \
-		((adev)->powerplay.pp_funcs->switch_power_profile(\
-			(adev)->powerplay.pp_handle, type, en))
+		if ((adev)->powerplay.pp_funcs->switch_power_profile != NULL) \
+			((adev)->powerplay.pp_funcs->switch_power_profile(\
+				(adev)->powerplay.pp_handle, type, en))
 
 #define amdgpu_dpm_set_clockgating_by_smu(adev, msg_id) \
 		((adev)->powerplay.pp_funcs->set_clockgating_by_smu(\
--
2.13.7

_______________________________________________
amd-gfx mailing list
amd-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/amd-gfx