Skip to contentSkip to conversation

Please backport commit 440cec4ca1c2 ("drm/amdgpu: Wait for bootloader after PSPv11 reset") to v6.16.y

PATCH Salvatore Bonaccorso 3 participants 4 messages Sep 14, 2025

Subsystems (3)
Stable branch DRM drivers Radeon and amdgpu DRM drivers
lore
Share & tools
lore

Salvatore Bonaccorso

author
#0 lore raw

Hi

In Debian we got the report in https://bugs.debian.org/1114806 that suspend to RAM fails (amdgpu driver hang) and Niklas Cathor was both able to bisect the issue down to 8345a71fc54b ("drm/amdgpu: Add more checks to PSP mailbox") (which was backported to 6.12.2 as well).

There is an upstream report as well at
https://gitlab.freedesktop.org/drm/amd/-/issues/4531 matching the issue and fixed by 440cec4ca1c2 ("drm/amdgpu: Wait for bootloader after PSPv11 reset").

Unfortunately the commit does not apply cleanly to 6.16.y as well as there were the changes around 9888f73679b7 ("drm/amdgpu: Add a noverbose flag to psp_wait_for").

Attached patch backports the commit due to this context changes, assuming it is not desirable to pick as well 9888f73679b7.

Does that looks good? If yes, can you please consider picking it up or the next 6.16.y stable series as well?

Regards,
Salvatore

From 6e2bbc123bf09d2f98008722c61d3907f54a0611 Mon Sep 17 00:00:00 2001
From: Lijo Lazar <[email protected]>
Date: Fri, 18 Jul 2025 18:50:58 +0530
Subject: [PATCH] drm/amdgpu: Wait for bootloader after PSPv11 reset

Commit 440cec4ca1c242d72e309a801995584a55af25c6 upstream.

Some PSPv11 SOCs take a longer time for PSP based mode-1 reset. Instead of checking for C2PMSG_33 status, add the callback wait_for_bootloader. Wait for bootloader to be back to steady state is already part of the generic mode-1 reset flow. Increase the retry count for bootloader wait and also fix the mask to prevent fake pass.

Fixes 8345a71fc54b ("drm/amdgpu: Add more checks to PSP mailbox")
Closes https://gitlab.freedesktop.org/drm/amd/-/issues/4531
Signed-off-by Lijo Lazar <[email protected]>
Reviewed-by Alex Deucher <[email protected]>
Signed-off-by Alex Deucher <[email protected]>
(cherry picked from commit 32f73741d6ee41fd5db8791c1163931e313d0fdc)

[Salvatore Bonaccorso: Backport for v6.16: Context changes for code before 9888f73679b7 ("drm/amdgpu: Add a noverbose flag to psp_wait_for")]

Signed-off-by Salvatore Bonaccorso <[email protected]>
---
 drivers/gpu/drm/amd/amdgpu/psp_v11_0.c | 19 ++++---------------
 1 file changed, 4 insertions(+), 15 deletions(-)
drivers/gpu/drm/amd/amdgpu/psp_v11_0.c +4 -15
index 1a4a26e6ffd2..4cd37fe96c6e 100644
@@ -149,13 +149,13 @@ static int psp_v11_0_wait_for_bootloader(struct psp_context *psp)
 	int ret;
 	int retry_loop;
 
-	for (retry_loop = 0; retry_loop < 10; retry_loop++) {
+	for (retry_loop = 0; retry_loop < 20; retry_loop++) {
 		/* Wait for bootloader to signify that is
 		    ready having bit 31 of C2PMSG_35 set to 1 */
 		ret = psp_wait_for(psp,
 				   SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
 				   0x80000000,
-				   0x80000000,
+				   0x8000FFFF,
 				   false);
 
 		if (ret == 0)

@@ -399,18 +399,6 @@ static int psp_v11_0_mode1_reset(struct psp_context *psp)
 
 	msleep(500);
 
-	offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_33);
-
-	ret = psp_wait_for(psp, offset, MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK,
-			   false);
-
-	if (ret) {
-		DRM_INFO("psp mode 1 reset failed!\n");
-		return -EINVAL;
-	}
-
-	DRM_INFO("psp mode1 reset succeed \n");
-
 	return 0;
 }
 

@@ -666,7 +654,8 @@ static const struct psp_funcs psp_v11_0_funcs = {
 	.ring_get_wptr = psp_v11_0_ring_get_wptr,
 	.ring_set_wptr = psp_v11_0_ring_set_wptr,
 	.load_usbc_pd_fw = psp_v11_0_load_usbc_pd_fw,
-	.read_usbc_pd_fw = psp_v11_0_read_usbc_pd_fw
+	.read_usbc_pd_fw = psp_v11_0_read_usbc_pd_fw,
+	.wait_for_bootloader = psp_v11_0_wait_for_bootloader
 };
 
 void psp_v11_0_set_psp_funcs(struct psp_context *psp)
Signature · 1 line
2.51.0
maintainer
#1 lore raw
Quoted 16 lines · Salvatore Bonaccorso

Hi

In Debian we got the report in https://bugs.debian.org/1114806 that suspend to RAM fails (amdgpu driver hang) and Niklas Cathor was both able to bisect the issue down to 8345a71fc54b ("drm/amdgpu: Add more checks to PSP mailbox") (which was backported to 6.12.2 as well).

There is an upstream report as well at
https://gitlab.freedesktop.org/drm/amd/-/issues/4531 matching the issue and fixed by 440cec4ca1c2 ("drm/amdgpu: Wait for bootloader after PSPv11 reset").

Unfortunately the commit does not apply cleanly to 6.16.y as well as there were the changes around 9888f73679b7 ("drm/amdgpu: Add a noverbose flag to psp_wait_for").

Quoted 5 lines

Attached patch backports the commit due to this context changes, assuming it is not desirable to pick as well 9888f73679b7.

Does that looks good? If yes, can you please consider picking it up or the next 6.16.y stable series as well?

I have a revert of the offending commit in the 6.16.y queue right now, as this was pointed out as causing a problem:

	https://lore.kernel.org/all/[email protected]/

so that should resolve this issue, right?

thanks,

greg k-h

author
#2 lore raw

Hi Greg,

[note fixing up my typo for the Debian bug, sorry about that]

Quoted 20 lines · Greg Kroah-Hartman
Quoted 19 lines · Salvatore Bonaccorso

Hi

In Debian we got the report in https://bugs.debian.org/1114806 that suspend to RAM fails (amdgpu driver hang) and Niklas Cathor was both able to bisect the issue down to 8345a71fc54b ("drm/amdgpu: Add more checks to PSP mailbox") (which was backported to 6.12.2 as well).

There is an upstream report as well at
https://gitlab.freedesktop.org/drm/amd/-/issues/4531 matching the issue and fixed by 440cec4ca1c2 ("drm/amdgpu: Wait for bootloader after PSPv11 reset").

Unfortunately the commit does not apply cleanly to 6.16.y as well as there were the changes around 9888f73679b7 ("drm/amdgpu: Add a noverbose flag to psp_wait_for").

Attached patch backports the commit due to this context changes, assuming it is not desirable to pick as well 9888f73679b7.

Quoted 7 lines
Quoted 2 lines

Does that looks good? If yes, can you please consider picking it up or the next 6.16.y stable series as well?

I have a revert of the offending commit in the 6.16.y queue right now, as this was pointed out as causing a problem:

	https://lore.kernel.org/all/[email protected]/

so that should resolve this issue, right?

Ah good, yes that should be equally fine (I missed the pending revert). Thanks a lot!

Regards,
Salvatore

#3 lore raw
In reply to Greg Kroah-Hartman

Hi Greg,

Quoted 20 lines · Greg Kroah-Hartman
Quoted 19 lines · Salvatore Bonaccorso

Hi

In Debian we got the report in https://bugs.debian.org/1114806 that suspend to RAM fails (amdgpu driver hang) and Niklas Cathor was both able to bisect the issue down to 8345a71fc54b ("drm/amdgpu: Add more checks to PSP mailbox") (which was backported to 6.12.2 as well).

There is an upstream report as well at
https://gitlab.freedesktop.org/drm/amd/-/issues/4531 matching the issue and fixed by 440cec4ca1c2 ("drm/amdgpu: Wait for bootloader after PSPv11 reset").

Unfortunately the commit does not apply cleanly to 6.16.y as well as there were the changes around 9888f73679b7 ("drm/amdgpu: Add a noverbose flag to psp_wait_for").

Attached patch backports the commit due to this context changes, assuming it is not desirable to pick as well 9888f73679b7.

Quoted 6 lines
Quoted 2 lines

Does that looks good? If yes, can you please consider picking it up or the next 6.16.y stable series as well?

I have a revert of the offending commit in the 6.16.y queue right now, as this was pointed out as causing a problem:

	https://lore.kernel.org/all/[email protected]/

so that should resolve this issue, right?

Yes, I can confirm that applying that revert to 6.16.y solves the issue on my machine as well.

cheers,
Niklas