Skip to contentSkip to conversation

Suspend to RAM fault in VT when resuming

PATCH Tim Gardner 2 participants 6 messages Mar 13, 2007 – Mar 19, 2007 Char and misc drivers

lore
Share & tools
lore
#0 lore raw

Pavel,

I've chased one of the 'Suspend to RAM' resume problems to a specific line in drivers/char/vt.c, see attached 2.6.21-rc3 diff with TRACE_RESUME() instrumentation. The macro scr_writew resolves to '*addr = val', which appears to be causing the problem. I've verified that the pointer is not NULL, but don't know if its really valid. Its pretty tough to tell what is happening, but on a Dell XPS it just hangs. A Dell Precision blinks the keyboard lights.

Since I don't know anything about video programming, I though perhaps you might have an opinion.

rtg

Signature · 1 line
Tim Gardner [email protected]
drivers/char/vt.c +3 -0
index c3f8e38..43a0904 100644
@@ -98,6 +98,7 @@
 #include <linux/pm.h>
 #include <linux/font.h>
 #include <linux/bitops.h>
+#include <linux/resume-trace.h>
 
 #include <asm/io.h>
 #include <asm/system.h>
@@ -2263,7 +2264,9 @@ static void vt_console_print(struct console *co, const char *b, unsigned count)
 			if (c == 10 || c == 13)
 				continue;
 		}
+TRACE_RESUME(0);
 		scr_writew((vc->vc_attr << 8) + c, (unsigned short *)vc->vc_pos);
+TRACE_RESUME(0);
 		cnt++;
 		if (myx == vc->vc_cols - 1) {
 			vc->vc_need_wrap = 1;
#1 lore raw

Hi1

Quoted 2 lines

I've chased one of the 'Suspend to RAM' resume problems to a specific line in drivers/char/vt.c, see attached 2.6.21-rc3 diff with

Has suspend/resume ever worked on that hardware?

Quoted 5 lines

TRACE_RESUME() instrumentation. The macro scr_writew resolves to '*addr = val', which appears to be causing the problem. I've verified that the pointer is not NULL, but don't know if its really valid. Its pretty tough to tell what is happening, but on a Dell XPS it just hangs. A Dell Precision blinks the keyboard lights.

It is possible that video is not initialized at that point, and that hardware goes seriously unhappy when you access non-existing vga. Does it resume ok when you completely disable video support?

Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#2 lore raw
Quoted 13 lines · Pavel Machek

Hi1

Quoted 2 lines

I've chased one of the 'Suspend to RAM' resume problems to a specific line in drivers/char/vt.c, see attached 2.6.21-rc3 diff with

Has suspend/resume ever worked on that hardware?

Quoted 5 lines

TRACE_RESUME() instrumentation. The macro scr_writew resolves to '*addr = val', which appears to be causing the problem. I've verified that the pointer is not NULL, but don't know if its really valid. Its pretty tough to tell what is happening, but on a Dell XPS it just hangs. A Dell Precision blinks the keyboard lights.

Quoted 8 lines

It is possible that video is not initialized at that point, and that hardware goes seriously unhappy when you access non-existing vga. Does it resume ok when you completely disable video support?

Pavel

Resume works on the Dell XPS with the Ubuntu Edgy release Ubuntu-2.6.17-10.25 (2.6.17 plus a zillion fixes). Ubuntu's git tree is rsync://rsync.kernel.org/pub/scm/linux/kernel/git/bcollins/ubuntu-2.6.git. Ubuntu-2.6.17-8.21 is the first version where resume works, but there are a boatload of changes in ACPI and SW suspend between that and the previous tag Ubuntu-2.6.17-7.20. I don't know what made SW suspend work, but even knowing that won't tell me what broke it again.

I've been avoiding the bisect process because it is quite time consuming on my slow machine, there is much branch weirdness, and I'm not that good with git. I thought if I narrowed the failure down to a small chunk of code in 2.6.21-rc3, then the answer might be obvious. No such luck, huh?

This crash behaves like the video memory space has become unmapped. Is that possible?

rtg

Signature · 1 line
Tim Gardner [email protected]
#3 lore raw
Quoted 26 lines · Tim Gardner
Quoted 13 lines · Pavel Machek

Hi1

Quoted 2 lines

I've chased one of the 'Suspend to RAM' resume problems to a specific line in drivers/char/vt.c, see attached 2.6.21-rc3 diff with

Has suspend/resume ever worked on that hardware?

Quoted 5 lines

TRACE_RESUME() instrumentation. The macro scr_writew resolves to '*addr = val', which appears to be causing the problem. I've verified that the pointer is not NULL, but don't know if its really valid. Its pretty tough to tell what is happening, but on a Dell XPS it just hangs. A Dell Precision blinks the keyboard lights.

Quoted 3 lines

It is possible that video is not initialized at that point, and that hardware goes seriously unhappy when you access non-existing vga. Does it resume ok when you completely disable video support?

Resume works on the Dell XPS with the Ubuntu Edgy release Ubuntu-2.6.17-10.25 (2.6.17 plus a zillion fixes). Ubuntu's git tree is rsync://rsync.kernel.org/pub/scm/linux/kernel/git/bcollins/ubuntu-2.6.git. Ubuntu-2.6.17-8.21 is the first version where resume works, but there are a boatload of changes in ACPI and SW suspend between that and the previous tag Ubuntu-2.6.17-7.20. I don't know what made SW suspend work, but even knowing that won't tell me what broke it again.

Quoted 7 lines

I've been avoiding the bisect process because it is quite time consuming on my slow machine, there is much branch weirdness, and I'm not that good with git. I thought if I narrowed the failure down to a small chunk of code in 2.6.21-rc3, then the answer might be obvious. No such luck, huh?

This crash behaves like the video memory space has become unmapped. Is that possible?

That seems improbable. It is physical ram, we do not unmap that..

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#4 lore raw
Quoted 29 lines · Pavel Machek
Quoted 28 lines · Tim Gardner
Quoted 6 lines · Pavel Machek

Hi1

Quoted 2 lines

I've chased one of the 'Suspend to RAM' resume problems to a specific line in drivers/char/vt.c, see attached 2.6.21-rc3 diff with

Has suspend/resume ever worked on that hardware?

Quoted 8 lines
Quoted 5 lines

TRACE_RESUME() instrumentation. The macro scr_writew resolves to '*addr = val', which appears to be causing the problem. I've verified that the pointer is not NULL, but don't know if its really valid. Its pretty tough to tell what is happening, but on a Dell XPS it just hangs. A Dell Precision blinks the keyboard lights.

It is possible that video is not initialized at that point, and that hardware goes seriously unhappy when you access non-existing vga. Does it resume ok when you completely disable video support?

Resume works on the Dell XPS with the Ubuntu Edgy release Ubuntu-2.6.17-10.25 (2.6.17 plus a zillion fixes). Ubuntu's git tree is rsync://rsync.kernel.org/pub/scm/linux/kernel/git/bcollins/ubuntu-2.6.git. Ubuntu-2.6.17-8.21 is the first version where resume works, but there are a boatload of changes in ACPI and SW suspend between that and the previous tag Ubuntu-2.6.17-7.20. I don't know what made SW suspend work, but even knowing that won't tell me what broke it again.

I've been avoiding the bisect process because it is quite time consuming on my slow machine, there is much branch weirdness, and I'm not that good with git. I thought if I narrowed the failure down to a small chunk of code in 2.6.21-rc3, then the answer might be obvious. No such luck, huh?

Quoted 5 lines
Quoted 2 lines

This crash behaves like the video memory space has become unmapped. Is that possible?

That seems improbable. It is physical ram, we do not unmap that..

The machine _has_ just come out of BIOS ACPI space. Perhaps PCI resources for video are hosed. All I can tell for sure is that this one line appears to cause some kind of crash. I'm working on finding some 1394 resources to see if I can nail this one.

Signature · 1 line
Tim Gardner [email protected]
#5 lore raw

Hi!

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

This crash behaves like the video memory space has become unmapped. Is that possible?

That seems improbable. It is physical ram, we do not unmap that..

The machine _has_ just come out of BIOS ACPI space. Perhaps PCI resources for video are hosed. All I can tell for sure is that this one

Yes, this is possible, AFAICT. I'm just saying that it is unlikely to be unmapped at pagetable level.
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html