mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Yazdan Ghalavand <yazdanghalavand@gmail.com>
To: nouveau@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org,
	Lyude Paul <lyude@redhat.com>, Danilo Krummrich <dakr@kernel.org>
Subject: [PATCH] drm/nouveau/bios/vmap: only read max2 if the header contains it
Date: Mon,  9 Nov 2026 10:41:02 +0300	[thread overview]
Message-ID: <20261109074102.11997-1-yazdanghalavand@gmail.com> (raw)

nvbios_vmap_parse() reads the id of the third "max voltage" entry
(max2) from offset 0xc of a version 0x20 voltage map table whenever
the table's *entry* length is at least 0xc. The entry length says
nothing about the size of the header, though. On a GeForce 710M
(GF117) the header is only 0xa bytes long while the entries are 0x26
bytes, so "max2" is actually read from byte 2 of the first voltage map
entry:

  header:  20 0a 26 12 01 c8 00 ff ff ff
  entry 0: 01 10 50 f8 0c 00 ...
                 ^^ read as max2 = 0x50

There is no vmap entry 0x50 (the table has 0x12 entries), so
nvkm_volt_map() falls back to interpreting the id as a voltage in
10 mV units, and nvkm_cstate_find_best() caps every cstate at 800 mV.
That is below the boot voltage (850 mV) and below what every cstate of
the board needs (900-950 mV), so no cstate is ever valid and setting
any pstate fails with -EINVAL.

Check the header length instead of the entry length before reading
max2.

Fermi reclocking is disabled upstream, so on this board the bug is
only visible with reclocking force-enabled, but the parser runs for
every GPU with a version 0x20 vmap table, including those where
reclocking is enabled.

Tested on the GeForce 710M with Fermi reclocking force-enabled locally:
without this patch, writing pstate 0f fails with -EINVAL; with it, the
pstate is accepted and the core clock goes to 774 MHz.

Fixes: 4a4555a7f1dd ("drm/nouveau/volt: Parse the max voltage map entries")
Signed-off-by: Yazdan Ghalavand <yazdanghalavand@gmail.com>
---
 drivers/gpu/drm/nouveau/nvkm/subdev/bios/vmap.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/bios/vmap.c b/drivers/gpu/drm/nouveau/nvkm/subdev/bios/vmap.c
index c228ca1..07b3b47 100644
--- a/drivers/gpu/drm/nouveau/nvkm/subdev/bios/vmap.c
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/bios/vmap.c
@@ -68,7 +68,7 @@ nvbios_vmap_parse(struct nvkm_bios *bios, u8 *ver, u8 *hdr, u8 *cnt, u8 *len,
 	case 0x20:
 		info->max0 = nvbios_rd08(bios, vmap + 0x7);
 		info->max1 = nvbios_rd08(bios, vmap + 0x8);
-		if (*len >= 0xc)
+		if (*hdr > 0xc)
 			info->max2 = nvbios_rd08(bios, vmap + 0xc);
 		else
 			info->max2 = 0xff;
-- 
2.56.0


                 reply	other threads:[~2026-11-09  7:41 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

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=20261109074102.11997-1-yazdanghalavand@gmail.com \
    --to=yazdanghalavand@gmail.com \
    --cc=dakr@kernel.org \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=lyude@redhat.com \
    --cc=nouveau@lists.freedesktop.org \
    /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

all inboxes | Powered by JetHome®