This is the mail archive of the systemtap@sourceware.org mailing list for the systemtap project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: dwarf info in modules.


On Mon, 12 Sep 2005, Frank Ch. Eigler wrote:

Tigran Aivazian <tigran_aivazian@symantec.com> writes:

upgrading elfutils to 0.115 solved the "no dwarf info found" but now I
get an oops:
[...]
0x100497fbdc8 0xffffffff801229da arch_copy_kprobe+0x16
(0xffffffffa01cdc20, 0x7fbffff314, 0x100497fbe84, 0x4, 0x100497fbf48)
[...]

I wonder if this is because the init function is in a special section (.init.text) of the .ko binary, which is unloaded/unmapped after initialization (in sys_init_module). By the time systemtap asks to insert a kprobe, there would be no "there" there.

I am not sure that this analysis is correct, because if I simply do modprobe microcode and place a probe at microcode_open then I get exactly the same oops (same backtrace). And microcode driver wasn't going anywhere, i.e. not unloaded.


Note that I didn't try to open the microcode device or do anything else other than stap microcode.stp which contained two lines pointing directly to microcode_open and NOT *@microcode.c

Also I forgot to mention that the problem with the trivial module occurs even if I point directly to a function in it (not ->init or ->exit functions) and not *@tmod.c globally.

Kind regards
Tigran


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]