ctfdwdiff: Simple shell script for testing the CTF encoder/decoder
First it gets a file with DWARF info, converts that to CTF and adds
a ".SUNW_ctf" ELF section to the file with DWARF info. Double debugging
foo! Pay for one, take two!
For tcp_input.o for instance, the result is:
[acme@doppio pahole]$ cat /tmp/tcp_input.o.diff
--- /tmp/tcp_input.o.ctf.c 2009-03-19 19:48:23.000000000 -0300
+++ /tmp/tcp_input.o.dwarf.c 2009-03-19 19:48:23.000000000 -0300
@@ -1811,7 +1811,7 @@
/* XXX 6 bytes hole, try to pack */
- void (*call)(const struct marker *, void *); /* 24 8 */
+ void (*call)(const struct marker *, void *, ...); /* 24 8 */
struct marker_probe_closure single; /* 32 16 */
struct marker_probe_closure * multi; /* 48 8 */
const char * tp_name; /* 56 8 */
[acme@doppio pahole]$
Now back to figuring out how to encode a VARARGS marker in CTF...
Ah, to use the script just do:
./ctfdwdiff foo.o
Some will crash, but we're working hard for fuller customer
satisfaction.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2009-03-19 23:50:09 +01:00
|
|
|
#!/bin/bash
|
|
|
|
|
|
2009-03-20 18:34:30 +01:00
|
|
|
results_dir=/tmp/ctfdwdiff
|
|
|
|
|
|
ctf_encoder: Convert DWARF functions to CTF
Finally we can use the Elf file already opened in dwarf_load, call
cu__for_each_cached_symtab_entry to iterate over the symtab entries,
this iterator will first call dwfl_module_getsymtab, that will do the
relocation that will allow us to go from the symtab address to the one
in the DWARF DW_TAG_subprogram tag DW_AT_low_pc attribute.
And voila, for a relatively complex single unit Linux kernel object
file, kernel/sched.o, we go from:
Just DWARF (gcc -g):
$ ls -la kernel/sched.o
1979011 kernel/sched.o
Then we run this to encode the CTF section:
$ pahole -Z kernel/sched.o
And get a file with both DWARF and CTF ELF sections:
$ ls -la kernel/sched.o
2019848 kernel/sched.o
We still need to encode the "OBJECTS", i.e. variables, but this
gets us from 1979011 (just DWARF) to:
$ strip--strip-debug kernel/sched.o
$ ls -la kernel/sched.o
-rw-rw-r-- 1 acme acme 507008 2009-03-30 23:01 kernel/sched.o
25% of the original size.
Of course we don't have inline expansion information, parameter names,
goto labels, etc, but should be good enough for most use cases.
See, without DWARF data, if we ask for it to use DWARF, nothing will be
printed, if we don't speficy the format, it will try first DWARF, it
will not find anything, it will try CTF:
$ pahole -F dwarf kernel/sched.o
$ pahole -C seq_operations kernel/sched.o
struct seq_operations {
void * (*start)(struct seq_file *, loff_t *); /* 0 8 */
void (*stop)(struct seq_file *, void *); /* 8 8 */
void * (*next)(struct seq_file *, void *, loff_t *); /* 16 8 */
int (*show)(struct seq_file *, void *); /* 24 8 */
/* size: 32, cachelines: 1, members: 4 */
/* last cacheline: 32 bytes */
};
$ $ pfunct -Vi -f schedule kernel/sched.o
void schedule(void);
{ /* low_pc=0xe01 */
}/* size: 83 */
$
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2009-03-31 03:54:29 +02:00
|
|
|
diff_tool() {
|
|
|
|
|
local tool=$1
|
|
|
|
|
local dwarf_options=$2
|
|
|
|
|
local ctf_options=$3
|
|
|
|
|
local obj=$4
|
|
|
|
|
|
2009-04-01 00:21:46 +02:00
|
|
|
diff=$results_dir/$obj.$tool.diff
|
|
|
|
|
ctf=$results_dir/$obj.$tool.ctf.c
|
|
|
|
|
dwarf=$results_dir/$obj.$tool.dwarf.c
|
ctf_encoder: Convert DWARF functions to CTF
Finally we can use the Elf file already opened in dwarf_load, call
cu__for_each_cached_symtab_entry to iterate over the symtab entries,
this iterator will first call dwfl_module_getsymtab, that will do the
relocation that will allow us to go from the symtab address to the one
in the DWARF DW_TAG_subprogram tag DW_AT_low_pc attribute.
And voila, for a relatively complex single unit Linux kernel object
file, kernel/sched.o, we go from:
Just DWARF (gcc -g):
$ ls -la kernel/sched.o
1979011 kernel/sched.o
Then we run this to encode the CTF section:
$ pahole -Z kernel/sched.o
And get a file with both DWARF and CTF ELF sections:
$ ls -la kernel/sched.o
2019848 kernel/sched.o
We still need to encode the "OBJECTS", i.e. variables, but this
gets us from 1979011 (just DWARF) to:
$ strip--strip-debug kernel/sched.o
$ ls -la kernel/sched.o
-rw-rw-r-- 1 acme acme 507008 2009-03-30 23:01 kernel/sched.o
25% of the original size.
Of course we don't have inline expansion information, parameter names,
goto labels, etc, but should be good enough for most use cases.
See, without DWARF data, if we ask for it to use DWARF, nothing will be
printed, if we don't speficy the format, it will try first DWARF, it
will not find anything, it will try CTF:
$ pahole -F dwarf kernel/sched.o
$ pahole -C seq_operations kernel/sched.o
struct seq_operations {
void * (*start)(struct seq_file *, loff_t *); /* 0 8 */
void (*stop)(struct seq_file *, void *); /* 8 8 */
void * (*next)(struct seq_file *, void *, loff_t *); /* 16 8 */
int (*show)(struct seq_file *, void *); /* 24 8 */
/* size: 32, cachelines: 1, members: 4 */
/* last cacheline: 32 bytes */
};
$ $ pfunct -Vi -f schedule kernel/sched.o
void schedule(void);
{ /* low_pc=0xe01 */
}/* size: 83 */
$
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2009-03-31 03:54:29 +02:00
|
|
|
$tool -F ctf $ctf_options $obj > $ctf
|
|
|
|
|
$tool -F dwarf $dwarf_options $obj > $dwarf
|
2009-03-20 18:34:30 +01:00
|
|
|
diff -up $dwarf $ctf > $diff
|
|
|
|
|
if [ -s $diff ] ; then
|
2009-04-01 00:21:46 +02:00
|
|
|
[ $# -gt 4 ] && vim $diff
|
2009-03-20 18:34:30 +01:00
|
|
|
exit 0
|
|
|
|
|
else
|
|
|
|
|
rm -f $diff $ctf $dwarf
|
|
|
|
|
fi
|
|
|
|
|
}
|
|
|
|
|
|
ctf_encoder: Convert DWARF functions to CTF
Finally we can use the Elf file already opened in dwarf_load, call
cu__for_each_cached_symtab_entry to iterate over the symtab entries,
this iterator will first call dwfl_module_getsymtab, that will do the
relocation that will allow us to go from the symtab address to the one
in the DWARF DW_TAG_subprogram tag DW_AT_low_pc attribute.
And voila, for a relatively complex single unit Linux kernel object
file, kernel/sched.o, we go from:
Just DWARF (gcc -g):
$ ls -la kernel/sched.o
1979011 kernel/sched.o
Then we run this to encode the CTF section:
$ pahole -Z kernel/sched.o
And get a file with both DWARF and CTF ELF sections:
$ ls -la kernel/sched.o
2019848 kernel/sched.o
We still need to encode the "OBJECTS", i.e. variables, but this
gets us from 1979011 (just DWARF) to:
$ strip--strip-debug kernel/sched.o
$ ls -la kernel/sched.o
-rw-rw-r-- 1 acme acme 507008 2009-03-30 23:01 kernel/sched.o
25% of the original size.
Of course we don't have inline expansion information, parameter names,
goto labels, etc, but should be good enough for most use cases.
See, without DWARF data, if we ask for it to use DWARF, nothing will be
printed, if we don't speficy the format, it will try first DWARF, it
will not find anything, it will try CTF:
$ pahole -F dwarf kernel/sched.o
$ pahole -C seq_operations kernel/sched.o
struct seq_operations {
void * (*start)(struct seq_file *, loff_t *); /* 0 8 */
void (*stop)(struct seq_file *, void *); /* 8 8 */
void * (*next)(struct seq_file *, void *, loff_t *); /* 16 8 */
int (*show)(struct seq_file *, void *); /* 24 8 */
/* size: 32, cachelines: 1, members: 4 */
/* last cacheline: 32 bytes */
};
$ $ pfunct -Vi -f schedule kernel/sched.o
void schedule(void);
{ /* low_pc=0xe01 */
}/* size: 83 */
$
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2009-03-31 03:54:29 +02:00
|
|
|
diff_one() {
|
|
|
|
|
local obj=$1
|
|
|
|
|
diff_tool "pahole" "--flat_arrays --show_private_classes --fixup_silly_bitfields" " " $obj $2
|
2009-04-01 00:21:46 +02:00
|
|
|
diff_tool "pfunct" "-V --no_parm_names" "-V" $obj $2
|
ctf_encoder: Convert DWARF functions to CTF
Finally we can use the Elf file already opened in dwarf_load, call
cu__for_each_cached_symtab_entry to iterate over the symtab entries,
this iterator will first call dwfl_module_getsymtab, that will do the
relocation that will allow us to go from the symtab address to the one
in the DWARF DW_TAG_subprogram tag DW_AT_low_pc attribute.
And voila, for a relatively complex single unit Linux kernel object
file, kernel/sched.o, we go from:
Just DWARF (gcc -g):
$ ls -la kernel/sched.o
1979011 kernel/sched.o
Then we run this to encode the CTF section:
$ pahole -Z kernel/sched.o
And get a file with both DWARF and CTF ELF sections:
$ ls -la kernel/sched.o
2019848 kernel/sched.o
We still need to encode the "OBJECTS", i.e. variables, but this
gets us from 1979011 (just DWARF) to:
$ strip--strip-debug kernel/sched.o
$ ls -la kernel/sched.o
-rw-rw-r-- 1 acme acme 507008 2009-03-30 23:01 kernel/sched.o
25% of the original size.
Of course we don't have inline expansion information, parameter names,
goto labels, etc, but should be good enough for most use cases.
See, without DWARF data, if we ask for it to use DWARF, nothing will be
printed, if we don't speficy the format, it will try first DWARF, it
will not find anything, it will try CTF:
$ pahole -F dwarf kernel/sched.o
$ pahole -C seq_operations kernel/sched.o
struct seq_operations {
void * (*start)(struct seq_file *, loff_t *); /* 0 8 */
void (*stop)(struct seq_file *, void *); /* 8 8 */
void * (*next)(struct seq_file *, void *, loff_t *); /* 16 8 */
int (*show)(struct seq_file *, void *); /* 24 8 */
/* size: 32, cachelines: 1, members: 4 */
/* last cacheline: 32 bytes */
};
$ $ pfunct -Vi -f schedule kernel/sched.o
void schedule(void);
{ /* low_pc=0xe01 */
}/* size: 83 */
$
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2009-03-31 03:54:29 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2009-03-22 17:06:41 +01:00
|
|
|
diff_dir() {
|
|
|
|
|
find . -type d | \
|
|
|
|
|
while read dir ; do
|
|
|
|
|
cd $dir
|
|
|
|
|
ls *.o 2> /dev/null |
|
|
|
|
|
while read obj ; do
|
|
|
|
|
ncus=$(readelf -wi $obj | grep DW_TAG_compile_unit | wc -l)
|
|
|
|
|
if [ $ncus -ne 1 ] ; then
|
|
|
|
|
continue
|
|
|
|
|
fi
|
|
|
|
|
echo $obj
|
ctf_encoder: Convert DWARF functions to CTF
Finally we can use the Elf file already opened in dwarf_load, call
cu__for_each_cached_symtab_entry to iterate over the symtab entries,
this iterator will first call dwfl_module_getsymtab, that will do the
relocation that will allow us to go from the symtab address to the one
in the DWARF DW_TAG_subprogram tag DW_AT_low_pc attribute.
And voila, for a relatively complex single unit Linux kernel object
file, kernel/sched.o, we go from:
Just DWARF (gcc -g):
$ ls -la kernel/sched.o
1979011 kernel/sched.o
Then we run this to encode the CTF section:
$ pahole -Z kernel/sched.o
And get a file with both DWARF and CTF ELF sections:
$ ls -la kernel/sched.o
2019848 kernel/sched.o
We still need to encode the "OBJECTS", i.e. variables, but this
gets us from 1979011 (just DWARF) to:
$ strip--strip-debug kernel/sched.o
$ ls -la kernel/sched.o
-rw-rw-r-- 1 acme acme 507008 2009-03-30 23:01 kernel/sched.o
25% of the original size.
Of course we don't have inline expansion information, parameter names,
goto labels, etc, but should be good enough for most use cases.
See, without DWARF data, if we ask for it to use DWARF, nothing will be
printed, if we don't speficy the format, it will try first DWARF, it
will not find anything, it will try CTF:
$ pahole -F dwarf kernel/sched.o
$ pahole -C seq_operations kernel/sched.o
struct seq_operations {
void * (*start)(struct seq_file *, loff_t *); /* 0 8 */
void (*stop)(struct seq_file *, void *); /* 8 8 */
void * (*next)(struct seq_file *, void *, loff_t *); /* 16 8 */
int (*show)(struct seq_file *, void *); /* 24 8 */
/* size: 32, cachelines: 1, members: 4 */
/* last cacheline: 32 bytes */
};
$ $ pfunct -Vi -f schedule kernel/sched.o
void schedule(void);
{ /* low_pc=0xe01 */
}/* size: 83 */
$
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2009-03-31 03:54:29 +02:00
|
|
|
pahole -Z $obj
|
2009-03-22 17:06:41 +01:00
|
|
|
diff_one $obj $1
|
|
|
|
|
done
|
|
|
|
|
cd - > /dev/null
|
|
|
|
|
done
|
|
|
|
|
}
|
|
|
|
|
|
2009-03-20 18:34:30 +01:00
|
|
|
rm -rf $results_dir
|
|
|
|
|
mkdir $results_dir
|
2009-03-22 17:06:41 +01:00
|
|
|
|
|
|
|
|
if [ $# -lt 2 ] ; then
|
|
|
|
|
diff_dir
|
|
|
|
|
else
|
|
|
|
|
diff_one $*
|
|
|
|
|
fi
|