Skip to contentSkip to conversation

2.4.20 unwind.c to handle multiple struct pt_regs

PATCH Applied Keith Owens 5 participants 10 messages Feb 20, 2003 – Mar 12, 2003

lore
Share & tools
lore
#0 lore raw
Quoted 7 lines · David Mosberger

Keith> This patch has been running inside SGI for 2 months. It Keith> handles kernel stacks with multiple struct pt_regs, as found Keith> while debugging the kernel. The debugging statements in the Keith> unwind code were awkward to use and inconsistent in what was Keith> being dumped and when, I ended up changing almost all the Keith> unwind debugging code while tracking down this problem.

Quoted 5 lines

The patch basically looks fine to me, except that I'd really like to prefer to have it split up into one patch that does the pt_regs fixes and one that does the debug enhancements. And while you're at it, please use something a bit more concise than UNW_DEBUG_PRINT (something like DPRINT or UNW_DPRINT should be sufficient).

Two patches follow. The first updates include/asm-ia64/unwind.h, arch/ia64/kernel/unwind_i.h, arch/ia64/kernel/unwind.c to handle multiple pt_regs on stack. The second only updates unwind.c, it goes over the first patch to clean up the debug code, using UNW_DPRINT().

asm-ia64/unwind.h +1 -0
diff include/asm-ia64/unwind.h
@@ -60,6 +60,7 @@ struct unw_frame_info {
 	unsigned long ip;		/* instruction pointer value */
 	unsigned long pr;		/* current predicate values */
 	unsigned long *cfm_loc;		/* cfm save location (or NULL) */
+	unsigned long pt;		/* struct pt_regs location */
 
 	struct task_struct *task;
 	struct switch_stack *sw;
ia64/kernel/unwind_i.h +2 -1
diff arch/ia64/kernel/unwind_i.h
@@ -137,7 +137,8 @@ enum unw_insn_opcode {
 	UNW_INSN_SETNAT_MEMSTK,		/* s[dst+1].nat.type = MEMSTK;
 					   s[dst+1].nat.off = *s.pri_unat - s[dst] */
 	UNW_INSN_SETNAT_TYPE,		/* s[dst+1].nat.type = val */
-	UNW_INSN_LOAD			/* s[dst] = *s[val] */
+	UNW_INSN_LOAD,			/* s[dst] = *s[val] */
+	UNW_INSN_MOVE_SCRATCH,		/* s[dst] = scratch reg "val" */
 };
 
 struct unw_insn {
ia64/kernel/unwind.c +41 -28

147 lines not on this page · raw message

ia64/kernel/unwind.c +112 -72

486 lines not on this page · raw message

#1 lore raw
Quoted 10 lines · Keith Owens
Quoted 5 lines

The patch basically looks fine to me, except that I'd really like to prefer to have it split up into one patch that does the pt_regs fixes and one that does the debug enhancements. And while you're at it, please use something a bit more concise than UNW_DEBUG_PRINT (something like DPRINT or UNW_DPRINT should be sufficient).

Two patches follow. The first updates include/asm-ia64/unwind.h, arch/ia64/kernel/unwind_i.h, arch/ia64/kernel/unwind.c to handle multiple pt_regs on stack. The second only updates unwind.c, it goes over the first patch to clean up the debug code, using UNW_DPRINT().

This patch was not included in the ia64 bundle for 2.4.21-pre5. Are there any problems with the patch? Do I need to send it again? It applies over 2.4.21-pre5 with a 2 line offset on the last hunk.

#2 lore raw
In reply to Keith Owens
Quoted 8 lines
Quoted 4 lines

Two patches follow. The first updates include/asm-ia64/unwind.h, arch/ia64/kernel/unwind_i.h, arch/ia64/kernel/unwind.c to handle multiple pt_regs on stack. The second only updates unwind.c, it goes over the first patch to clean up the debug code, using UNW_DPRINT().

This patch was not included in the ia64 bundle for 2.4.21-pre5. Are there any problems with the patch? Do I need to send it again? It applies over 2.4.21-pre5 with a 2 line offset on the last hunk.

Wow, somebody is actually looking at the BK tree :-)

Your patches are still in my "todo" list. I haven't looked at it closely, but my intent is to put it in before the real 2.4.21 patch.

Bjorn

#3 lore raw
In reply to Keith Owens
Quoted 1 line · Bjorn Helgaas

Wow, somebody is actually looking at the BK tree :-)

I do not use BK. Because I work on PRCS I am not comfortable with using BK, there is a possible conflict with the BK 'no competitor' clause. I was looking at Marcelo's 2.4.21-pre5 patch.

Quoted 2 lines

Your patches are still in my "todo" list. I haven't looked at it closely, but my intent is to put it in before the real 2.4.21 patch.

Thanks.

#4 lore raw
In reply to Keith Owens
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line · Keith Owens

Keith> Two patches follow. The first updates include/asm-ia64/unwind.h, Keith> arch/ia64/kernel/unwind_i.h, arch/ia64/kernel/unwind.c to handle Keith> multiple pt_regs on stack. The second only updates unwind.c, it goes Keith> over the first patch to clean up the debug code, using UNW_DPRINT().

I applied most of these two patches now. I omitted the portions where you turned error messages into debug statements. Also, I'm going to change dump_info_pt() into something more reasonable.

	--david
#5 lore raw
In reply to Keith Owens
Quoted 10 lines · David Mosberger
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line · Keith Owens

Keith> Two patches follow. The first updates include/asm-ia64/unwind.h, Keith> arch/ia64/kernel/unwind_i.h, arch/ia64/kernel/unwind.c to handle Keith> multiple pt_regs on stack. The second only updates unwind.c, it goes Keith> over the first patch to clean up the debug code, using UNW_DPRINT().

I applied most of these two patches now. I omitted the portions where you turned error messages into debug statements. Also, I'm going to change dump_info_pt() into something more reasonable.

I applied these patches to 2.4 also:

[email protected], 2003-03-07 12:02:17-07:00, [email protected]
ia64: In kernel unwinder, replace dump_info_pt() with get_scratch_regs()
and reformat to make it fit in 100 columns.

[email protected], 2003-03-07 12:02:07-07:00, [email protected]
ia64; Improve debug output from kernel unwinder. Based on patch by
Keith Owens. (Ported to 2.4 by Bjorn Helgaas).

[email protected], 2003-03-07 12:01:27-07:00, [email protected]
[PATCH] ia64: fix scratch-regs handling in kernel unwinder

This patch has been running inside SGI for 2 months. It handles kernel stacks with multiple struct pt_regs, as found while debugging the kernel.

#6 lore raw
In reply to Keith Owens

[Sorry if this is a duplicate; I sent it yesterday, but it bounced due to an HP mail problem.]

Quoted 7 lines · Keith Owens
Quoted 1 line · Bjorn Helgaas

Wow, somebody is actually looking at the BK tree :-)

I do not use BK. Because I work on PRCS I am not comfortable with using BK, there is a possible conflict with the BK 'no competitor' clause. I was looking at Marcelo's 2.4.21-pre5 patch.

Yeah, that makes sense. BK is extremely convenient for me, because it makes it so easy to track what Marcelo and David are doing, but I certainly don't want to force that on anybody else.

It feels like the "right thing" to make plain old patches available whenever I publish changes via BK, and I'm willing to do that if there's any interest in it.

Bjorn

#7 lore raw
In reply to Keith Owens
Quoted 3 lines · Bjorn Helgaas

It feels like the "right thing" to make plain old patches available whenever I publish changes via BK, and I'm willing to do that if there's any interest in it.

Yes please. http://ftp.kernel.org/pub/linux/kernel/v2.[45]/testing/cset/
contains bk changesets converted to plain patches, we need the same for ia64 trees for people who can/will not use bk. I think Dave Jones set up the patche system.

#8 lore raw
In reply to Keith Owens
Quoted 10 lines · Keith Owens
Quoted 3 lines · Bjorn Helgaas

It feels like the "right thing" to make plain old patches available whenever I publish changes via BK, and I'm willing to do that if there's any interest in it.

Yes please. http://ftp.kernel.org/pub/linux/kernel/v2.[45]/testing/cset/
contains bk changesets converted to plain patches, we need the same for ia64 trees for people who can/will not use bk. I think Dave Jones set up the patche system.

Dave Woodhouse, actually... Dave, can you set this up for other trees?

Signature · 3 lines
"It's not Hollywood.  War is real, war is primarily not about defeat or
victory, it is about death.  I've seen thousands and thousands of dead bodies.
Do you think I want to have an academic debate on this subject?" -- Robert Fisk
#9 lore raw

Re: [Linux-ia64] Re: [patch] 2.4.20 unwind.c to handle multiple

In reply to Keith Owens
Quoted 7 lines · Matthew Wilcox
Quoted 6 lines · Keith Owens
Quoted 3 lines · Bjorn Helgaas

It feels like the "right thing" to make plain old patches available whenever I publish changes via BK, and I'm willing to do that if there's any interest in it.

Quoted 6 lines
Quoted 4 lines

Yes please. http://ftp.kernel.org/pub/linux/kernel/v2.[45]/testing/cset/
contains bk changesets converted to plain patches, we need the same for ia64 trees for people who can/will not use bk. I think Dave Jones set up the patche system.

Dave Woodhouse, actually... Dave, can you set this up for other trees?

export CVSROOT=:pserver:[email protected]:/home/cvs grep -q $CVSROOT ~/.cvspass || echo "$CVSROOT Ay=0=h<Z" >> ~/.cvspass cvs co bkexport

Signature · 1 line
dwmw2