public inbox for pve-devel@lists.proxmox.com
 help / color / mirror / Atom feed
From: Fiona Ebner <f.ebner@proxmox.com>
To: pve-devel@lists.proxmox.com
Subject: [pve-devel] [PATCH kernel 2/2] cherry-pick potential fix for NULL pointer deref with AMD Arcturus GPU during boot
Date: Wed, 10 Jul 2024 13:37:09 +0200	[thread overview]
Message-ID: <20240710113709.1415909-2-f.ebner@proxmox.com> (raw)
In-Reply-To: <20240710113709.1415909-1-f.ebner@proxmox.com>

The issue was reported in the enterprise support and is handled by
Alexander Zeidler. It has the following trace [0] and causes an issue
with the networking down the line, because 'udevadm settle' would time
out. The customer reported that mainline kernel 6.9.3 booted fine.
Looking at the new commits, this one stood out, as it heavily modifies
the arcturus_get_power_limit() function. While not tagged for stable,
it seems straightforward enough and has a good chance to fix the
issue.

[0]:

> Jul 09 07:34:59 proxmox kernel: BUG: kernel NULL pointer dereference, address: 000000000000000f
> Jul 09 07:34:59 proxmox kernel: #PF: supervisor read access in kernel mode
> Jul 09 07:34:59 proxmox kernel: #PF: error_code(0x0000) - not-present page
> Jul 09 07:34:59 proxmox kernel: PGD 0 P4D 0
> Jul 09 07:34:59 proxmox kernel: Oops: 0000 [#1] PREEMPT SMP NOPTI
> Jul 09 07:34:59 proxmox kernel: CPU: 0 PID: 9 Comm: kworker/0:1 Tainted: P           O       6.8.8-2-pve #1
> Jul 09 07:34:59 proxmox kernel: Hardware name: Supermicro AS -4124GS-TNR-03-EB004/H12DSG-O-CPU, BIOS 2.7 09/21/2023
> Jul 09 07:34:59 proxmox kernel: Workqueue: events work_for_cpu_fn
> Jul 09 07:34:59 proxmox kernel: RIP: 0010:arcturus_get_power_limit+0xb5/0x1b0 [amdgpu]
> Jul 09 07:34:59 proxmox kernel: Code: 24 48 85 d2 74 05 8b 45 cc 89 02 4d 85 ff 74 38 44 0f b6 a3 b8 06 00 00 41 80 fc 01 0f 87 81 d7 3d 00 48 8b 45 b0 41 83 e4 01 <0f> b6 40 0f 75 10 84 c0 74 14 45 8b bf 86 01 00 00 45 31 e4 eb 0e
> Jul 09 07:34:59 proxmox kernel: RSP: 0018:ffffaa42c029fc38 EFLAGS: 00010246
> Jul 09 07:34:59 proxmox kernel: RAX: 0000000000000000 RBX: ffff8d803362b000 RCX: 0000000000000000
> Jul 09 07:34:59 proxmox kernel: RDX: ffff8d803362b6c0 RSI: 0000000000000000 RDI: 0000000000000000
> Jul 09 07:34:59 proxmox kernel: RBP: ffffaa42c029fc88 R08: 0000000000000000 R09: ffffffffc177e1f0
> Jul 09 07:34:59 proxmox kernel: R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
> Jul 09 07:34:59 proxmox kernel: R13: ffff8d803362b6c8 R14: ffff8d803362b6c4 R15: ffff8d80424a1014
> Jul 09 07:34:59 proxmox kernel: FS:  0000000000000000(0000) GS:ffff8e7f0ae00000(0000) knlGS:0000000000000000
> Jul 09 07:34:59 proxmox kernel: CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> Jul 09 07:34:59 proxmox kernel: CR2: 000000000000000f CR3: 0000006b24a36003 CR4: 0000000000f70ef0
> Jul 09 07:34:59 proxmox kernel: PKRU: 55555554
> Jul 09 07:34:59 proxmox kernel: Call Trace:
> Jul 09 07:34:59 proxmox kernel:  <TASK>
> Jul 09 07:34:59 proxmox kernel:  ? show_regs+0x6d/0x80
> Jul 09 07:34:59 proxmox kernel:  ? __die+0x24/0x80
> Jul 09 07:34:59 proxmox kernel:  ? page_fault_oops+0x176/0x500
> Jul 09 07:34:59 proxmox kernel:  ? do_user_addr_fault+0x2f9/0x6b0
> Jul 09 07:34:59 proxmox kernel:  ? exc_page_fault+0x83/0x1b0
> Jul 09 07:34:59 proxmox kernel:  ? asm_exc_page_fault+0x27/0x30
> Jul 09 07:34:59 proxmox kernel:  ? __pfx_arcturus_get_power_limit+0x10/0x10 [amdgpu]
> Jul 09 07:34:59 proxmox kernel:  ? arcturus_get_power_limit+0xb5/0x1b0 [amdgpu]
> Jul 09 07:34:59 proxmox kernel:  ? arcturus_get_power_limit+0x62/0x1b0 [amdgpu]
> Jul 09 07:34:59 proxmox kernel:  smu_late_init+0x16f/0x4d0 [amdgpu]
> Jul 09 07:34:59 proxmox kernel:  amdgpu_device_ip_late_init+0x68/0x2a0 [amdgpu]
> Jul 09 07:34:59 proxmox kernel:  amdgpu_device_init+0x242d/0x26e0 [amdgpu]
> Jul 09 07:34:59 proxmox kernel:  ? srso_alias_return_thunk+0x5/0xfbef5
> Jul 09 07:34:59 proxmox kernel:  amdgpu_driver_load_kms+0x1a/0x1c0 [amdgpu]
> Jul 09 07:34:59 proxmox kernel:  amdgpu_pci_probe+0x195/0x520 [amdgpu]
> Jul 09 07:34:59 proxmox kernel:  local_pci_probe+0x47/0xb0
> Jul 09 07:34:59 proxmox kernel:  work_for_cpu_fn+0x1a/0x30
> Jul 09 07:34:59 proxmox kernel:  process_one_work+0x16d/0x350
> Jul 09 07:34:59 proxmox kernel:  worker_thread+0x306/0x440
> Jul 09 07:34:59 proxmox kernel:  ? __pfx_worker_thread+0x10/0x10
> Jul 09 07:34:59 proxmox kernel:  kthread+0xf2/0x120
> Jul 09 07:34:59 proxmox kernel:  ? __pfx_kthread+0x10/0x10
> Jul 09 07:34:59 proxmox kernel:  ret_from_fork+0x47/0x70
> Jul 09 07:34:59 proxmox kernel:  ? __pfx_kthread+0x10/0x10
> Jul 09 07:34:59 proxmox kernel:  ret_from_fork_asm+0x1b/0x30
> Jul 09 07:34:59 proxmox kernel:  </TASK>

Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
---
 ...pu-pm-Don-t-use-OD-table-on-Arcturus.patch | 69 +++++++++++++++++++
 1 file changed, 69 insertions(+)
 create mode 100644 patches/kernel/0019-drm-amdgpu-pm-Don-t-use-OD-table-on-Arcturus.patch

diff --git a/patches/kernel/0019-drm-amdgpu-pm-Don-t-use-OD-table-on-Arcturus.patch b/patches/kernel/0019-drm-amdgpu-pm-Don-t-use-OD-table-on-Arcturus.patch
new file mode 100644
index 0000000..cd88e43
--- /dev/null
+++ b/patches/kernel/0019-drm-amdgpu-pm-Don-t-use-OD-table-on-Arcturus.patch
@@ -0,0 +1,69 @@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: Ma Jun <Jun.Ma2@amd.com>
+Date: Tue, 19 Mar 2024 11:02:29 +0800
+Subject: [PATCH] drm/amdgpu/pm: Don't use OD table on Arcturus
+
+OD is not supported on Arcturus, so the OD table
+should not be used.
+
+Signed-off-by: Ma Jun <Jun.Ma2@amd.com>
+Acked-by: Alex Deucher <alexander.deucher@amd.com>
+Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
+(cherry picked from commit bc55c344b06f7e6f99eb92d393ff0a84c1532514)
+Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
+---
+ .../gpu/drm/amd/pm/swsmu/smu11/arcturus_ppt.c | 33 +++----------------
+ 1 file changed, 5 insertions(+), 28 deletions(-)
+
+diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/arcturus_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/arcturus_ppt.c
+index 40ba7227cca5..0c2d04f978ac 100644
+--- a/drivers/gpu/drm/amd/pm/swsmu/smu11/arcturus_ppt.c
++++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/arcturus_ppt.c
+@@ -1283,11 +1283,8 @@ static int arcturus_get_power_limit(struct smu_context *smu,
+ 					uint32_t *max_power_limit,
+ 					uint32_t *min_power_limit)
+ {
+-	struct smu_11_0_powerplay_table *powerplay_table =
+-		(struct smu_11_0_powerplay_table *)smu->smu_table.power_play_table;
+-	struct smu_11_0_overdrive_table *od_settings = smu->od_settings;
+ 	PPTable_t *pptable = smu->smu_table.driver_pptable;
+-	uint32_t power_limit, od_percent_upper = 0, od_percent_lower = 0;
++	uint32_t power_limit;
+ 
+ 	if (smu_v11_0_get_current_power_limit(smu, &power_limit)) {
+ 		/* the last hope to figure out the ppt limit */
+@@ -1303,30 +1300,10 @@ static int arcturus_get_power_limit(struct smu_context *smu,
+ 		*current_power_limit = power_limit;
+ 	if (default_power_limit)
+ 		*default_power_limit = power_limit;
+-
+-	if (powerplay_table) {
+-		if (smu->od_enabled &&
+-				od_settings->cap[SMU_11_0_ODCAP_POWER_LIMIT]) {
+-			od_percent_upper = le32_to_cpu(powerplay_table->overdrive_table.max[SMU_11_0_ODSETTING_POWERPERCENTAGE]);
+-			od_percent_lower = le32_to_cpu(powerplay_table->overdrive_table.min[SMU_11_0_ODSETTING_POWERPERCENTAGE]);
+-		} else if (od_settings->cap[SMU_11_0_ODCAP_POWER_LIMIT]) {
+-			od_percent_upper = 0;
+-			od_percent_lower = le32_to_cpu(powerplay_table->overdrive_table.min[SMU_11_0_ODSETTING_POWERPERCENTAGE]);
+-		}
+-	}
+-
+-	dev_dbg(smu->adev->dev, "od percent upper:%d, od percent lower:%d (default power: %d)\n",
+-							od_percent_upper, od_percent_lower, power_limit);
+-
+-	if (max_power_limit) {
+-		*max_power_limit = power_limit * (100 + od_percent_upper);
+-		*max_power_limit /= 100;
+-	}
+-
+-	if (min_power_limit) {
+-		*min_power_limit = power_limit * (100 - od_percent_lower);
+-		*min_power_limit /= 100;
+-	}
++	if (max_power_limit)
++		*max_power_limit = power_limit;
++	if (min_power_limit)
++		*min_power_limit = power_limit;
+ 
+ 	return 0;
+ }
-- 
2.39.2



_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel


  reply	other threads:[~2024-07-10 11:36 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-07-10 11:37 [pve-devel] [PATCH kernel 1/2] add fix for CIFS client memory leak Fiona Ebner
2024-07-10 11:37 ` Fiona Ebner [this message]
2024-07-16  9:41   ` [pve-devel] [PATCH kernel 2/2] cherry-pick potential fix for NULL pointer deref with AMD Arcturus GPU during boot Alexander Zeidler
2024-07-12 14:54 ` [pve-devel] applied-series: [PATCH kernel 1/2] add fix for CIFS client memory leak Thomas Lamprecht

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20240710113709.1415909-2-f.ebner@proxmox.com \
    --to=f.ebner@proxmox.com \
    --cc=pve-devel@lists.proxmox.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
Service provided by Proxmox Server Solutions GmbH | Privacy | Legal