When I use markup_oops.pl parse a x8664 oops, I got:
objdump: --start-address: bad number: NaN
No matching code found
This is because:
main::(./m.pl:228): open(FILE, "objdump -dS --adjust-vma=$vmaoffset --start-address=$decodestart --stop-address=$decodestop $filename |") || die "Cannot start objdump";
DB<3> p $decodestart
NaN
This NaN is from:
main::(./m.pl:176): my $decodestart = Math::BigInt->from_hex("0x$target") - Math::BigInt->from_hex("0x$func_offset");
DB<2> p $func_offset
0x175
There is already a "0x" in $func_offset, another 0x makes it a NaN.
The $func_offset is from line:
if ($line =~ /RIP: 0010:\[\<[0-9a-f]+\>\] \[\<[0-9a-f]+\>\] ([a-zA-Z0-9\_]+)\+(0x[0-9a-f]+)\/0x[a-f0-9]/) {
$function = $1;
$func_offset = $2;
}
I make a patch to change "(0x[0-9a-f]+)\/0x[a-f0-9]/)" to "0x([0-9a-f]+)\/0x[a-f0-9]/)".
Signed-off-by: Hui Zhu <teawater@gmail.com>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Michal Marek <mmarek@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When I try to use markup_oops.pl in x86, I always get:
cat 1 | perl markup_oops.pl ./vmlinux
objdump: --start-address: bad number: NaN
No matching code found
This is because in line:
if ($line =~ /EIP is at ([a-zA-Z0-9\_]+)\+0x([0-9a-f]+)\/[a-f0-9]/) {
$function = $1;
$func_offset = $2;
}
$func_offset will get a number like "0x2"
But in follow code:
my $decodestart = Math::BigInt->from_hex("0x$target") -
Math::BigInt->from_hex("0x$func_offset");
It add other ox to ox2. Then this value will be set to NaN.
So I made a small patch to fix it.
Signed-off-by: Hui Zhu <teawater@gmail.com>
Acked-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When EIP is at a module having an underscore in its name, the current code
fails to find it because the module filenames has '-' instead of '_'. Use
modinfo for a better path finding.
Signed-off-by: Ozan Çaglayan <ozan@pardus.org.tr>
Acked-by: WANG Cong <xiyou.wangcong@gmail.com>
Cc: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
A 32-bit perl can't handle 64-bit addresses without using the BigInt
package.
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Acked-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
An oops dump also contains the register values.
This patch parses these for (32 bit) x86, and then annotates the
disassembly with these values; this helps in analysis of the oops by the
developer, for example, NULL pointer or other pointer bugs show up clearly
this way.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
There has been some light flamewar on lkml about decoding oopses
in modules (as part of the crashdump flamewar).
Now this isn't rocket science, just the markup_oops.pl script
cheaped out and didn't handle modules. But really; a flamewar
all about that?? What happened to C++ in the kernel or reading
files from inside the kernel?
This patch adds module support to markup_oops.pl; it's not the
most pretty perl but it works for my testcases...
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We're struggling all the time to figure out where the code came from that
oopsed.. The script below (a adaption from a script used by
kerneloops.org) can help developers quite a bit, at least for non-module
cases.
It works and looks like this:
[/home/arjan/linux]$ dmesg | perl scripts/markup_oops.pl vmlinux
{
struct agp_memory *memory;
memory = agp_allocate_memory(agp_bridge, pg_count, type);
c055c10f: 89 c2 mov %eax,%edx
if (memory == NULL)
c055c111: 74 19 je c055c12c <agp_allocate_memory_wrap+0x30>
/* This function must only be called when current_controller != NULL */
static void agp_insert_into_pool(struct agp_memory * temp)
{
struct agp_memory *prev;
prev = agp_fe.current_controller->pool;
c055c113: a1 ec dc 8f c0 mov 0xc08fdcec,%eax
*c055c118: 8b 40 10 mov 0x10(%eax),%eax <----- faulting instruction
if (prev != NULL) {
c055c11b: 85 c0 test %eax,%eax
c055c11d: 74 05 je c055c124 <agp_allocate_memory_wrap+0x28>
prev->prev = temp;
c055c11f: 89 50 04 mov %edx,0x4(%eax)
temp->next = prev;
c055c122: 89 02 mov %eax,(%edx)
}
agp_fe.current_controller->pool = temp;
c055c124: a1 ec dc 8f c0 mov 0xc08fdcec,%eax
c055c129: 89 50 10 mov %edx,0x10(%eax)
if (memory == NULL)
return NULL;
agp_insert_into_pool(memory);
so in this case, we faulted while dereferencing agp_fe.current_controller
pointer, and we get to see exactly which function and line it affects...
Personally I find this very useful, and I can see value for having this
script in the kernel for more-than-just-me to use.
Caveats:
* It only works for oopses not-in-modules
* It only works nicely for kernels compiled with CONFIG_DEBUG_INFO
* It's not very fast.
* It only works on x86
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Hui Zhu