Linux kernel -stable discussions
 help / color / mirror / Atom feed
From: Rand Deeb <rand.sec96@gmail.com>
To: stable@vger.kernel.org
Cc: Nathan Chancellor <nathan@kernel.org>,
	Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>,
	lvc-project@linuxtesting.org, deeb.rand@confident.ru,
	shda@confident.ru, kernel test robot <lkp@intel.com>,
	Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>,
	Rand Deeb <rand.sec96@gmail.com>
Subject: [PATCH 6.1.y] drm/msm/dp: Avoid division by zero in msm_dp_ctrl_config_msa()
Date: Fri, 20 Nov 2026 12:51:01 +0300	[thread overview]
Message-ID: <20261120095101.76-1-rand.sec96@gmail.com> (raw)

From: Nathan Chancellor <nathan@kernel.org>

[ Upstream commit f185076da44c774241a16a82a7773ece3c1c607b ]

An (admittedly problematic) optimization change in LLVM 20 [1] turns
known division by zero into the equivalent of __builtin_unreachable(),
which invokes undefined behavior if it is encountered in a control flow
graph, destroying code generation. When compile testing for x86_64,
objtool flags an instance of this optimization triggering in
msm_dp_ctrl_config_msa(), inlined into msm_dp_ctrl_on_stream():

  drivers/gpu/drm/msm/msm.o: warning: objtool: msm_dp_ctrl_on_stream(): unexpected end of section .text.msm_dp_ctrl_on_stream

The zero division happens if the else branch in the first if statement
in msm_dp_ctrl_config_msa() is taken because pixel_div is initialized to
zero and it is not possible for LLVM to eliminate the else branch since
rate is still not known after inlining into msm_dp_ctrl_on_stream().

Transform the if statements into a switch statement with a default case
with the existing error print and an early return to avoid the invalid
division. Add a comment to note this helps the compiler, even though the
case is known to be unreachable. With this, pixel_dev's default zero
initialization can be dropped, as it is dead with this change.

Fixes: c943b4948b58 ("drm/msm/dp: add displayPort driver support")
Link: https://github.com/llvm/llvm-project/commit/37932643abab699e8bb1def08b7eb4eae7ff1448 [1]
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202601081959.9UVJEOfP-lkp@intel.com/
Suggested-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/698355/
Link: https://lore.kernel.org/r/20260113-drm-msm-dp_ctrl-avoid-zero-div-v2-1-f1aa67bf6e8e@kernel.org
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
[ The function is still dp_catalog_ctrl_config_msa() in dp_catalog.c in
  this tree; same change applied there. ]
Signed-off-by: Rand Deeb <rand.sec96@gmail.com>
---
 drivers/gpu/drm/msm/dp/dp_catalog.c | 24 ++++++++++++++++++------
 1 file changed, 18 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/msm/dp/dp_catalog.c b/drivers/gpu/drm/msm/dp/dp_catalog.c
index 421391755..61f9cb018 100644
--- a/drivers/gpu/drm/msm/dp/dp_catalog.c
+++ b/drivers/gpu/drm/msm/dp/dp_catalog.c
@@ -420,7 +420,7 @@ void dp_catalog_ctrl_config_msa(struct dp_catalog *dp_catalog,
 					bool fixed_nvid)
 {
 	u32 pixel_m, pixel_n;
-	u32 mvid, nvid, pixel_div = 0, dispcc_input_rate;
+	u32 mvid, nvid, pixel_div, dispcc_input_rate;
 	u32 const nvid_fixed = DP_LINK_CONSTANT_N_VALUE;
 	u32 const link_rate_hbr2 = 540000;
 	u32 const link_rate_hbr3 = 810000;
@@ -429,14 +429,26 @@ void dp_catalog_ctrl_config_msa(struct dp_catalog *dp_catalog,
 	struct dp_catalog_private *catalog = container_of(dp_catalog,
 				struct dp_catalog_private, dp_catalog);
 
-	if (rate == link_rate_hbr3)
+	switch (rate) {
+	case link_rate_hbr3:
 		pixel_div = 6;
-	else if (rate == 162000 || rate == 270000)
-		pixel_div = 2;
-	else if (rate == link_rate_hbr2)
+		break;
+	case link_rate_hbr2:
 		pixel_div = 4;
-	else
+		break;
+	case 162000:
+	case 270000:
+		pixel_div = 2;
+		break;
+	default:
+		/*
+		 * This cannot be reached but the compiler is not able to know
+		 * that statically so return early to avoid a possibly invalid
+		 * division.
+		 */
 		DRM_ERROR("Invalid pixel mux divider\n");
+		return;
+	}
 
 	dispcc_input_rate = (rate * 10) / pixel_div;
 

base-commit: c75b0cf0982c00cddfd88d1acf3872f79b40980c
-- 
2.39.5


                 reply	other threads:[~2026-11-20  9:51 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=20261120095101.76-1-rand.sec96@gmail.com \
    --to=rand.sec96@gmail.com \
    --cc=deeb.rand@confident.ru \
    --cc=dmitry.baryshkov@oss.qualcomm.com \
    --cc=konrad.dybcio@oss.qualcomm.com \
    --cc=lkp@intel.com \
    --cc=lvc-project@linuxtesting.org \
    --cc=nathan@kernel.org \
    --cc=shda@confident.ru \
    --cc=stable@vger.kernel.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