-
Notifications
You must be signed in to change notification settings - Fork 2
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Replay diverge on modified log: bug make-abr2gbr #7
Comments
Log file: hp07:/persist/need-diagnose/make-abr2gbr/out.132.log |
thanks. can you also say where we can look at the original and modified On 07/23/2011 06:06 PM, chiache wrote:
|
original log: hp07:/persist/need-diagnose/make-abr2gbr/out.log other logs: Log file: hp07:/persist/need-diagnose/make-abr2gbr/out.*.log |
Output: hp07:/persist/need-diagnose/make-abr2gbr/result |
please provide the following (and only this) infromation:
pleaes use the latest version of the code to generate these. thanks ! On 07/26/2011 10:50 AM, chiache wrote:
|
Update: original log: hp05:/persist/make-abr2gbr/out.log backtrace: Backtrace (100 events): |
thanks for the additional info. unfortunately, there is not enough info in the log to figure out can you please generate a longer log ? On 07/29/2011 12:28 PM, chiache wrote:
|
chia-che, nico and I investigated the makefile divergence problem. we reached the conclusion that the divergence for RACE 5 is entirely in other words, we were chasing the wrong target ! we identified the pair of system calls in the log which are the root we then ran 'racepro show-races -C 200 ..." and it generated a total of in particular, RACE 98 is the one that should trigger the problem. (see the entire output below). everything is in /tmp/orenl on hp05, and the generated logs and the we didn't try to replay them -> that's for you. please see if the bug is produced by that race. thanks ! On 07/29/2011 12:28 PM, chiache wrote:
[...] ---- output file ---- root@hp05:/tmp/orenl# racepro -v show-races -c 200 -o blah/blah out.log INFO:root:resource 757 has too many events (217); skipRESOURCE found 125 potential racesRACE 1: pid 11 #179 [stat64()=-2] -> pid 9 #126 [mkdir()=0] RACE 125: pid 11 #213 [stat64()=-2] -> pid 9 #126 [mkdir()=0]EXIT-WAIT found 2 potential racesRACE 126: pid 7 #51 [exit_group()=0] -> pid 6 #197 [wait4()=8] -> pid 8 #75 [exit_group()=0] RACE 127: pid 10 #162 [exit_group()=0] -> pid 6 #209 [wait4()=7] -> pid 7 #51 [exit_group()=0]SIGNAL found 0 potential racesGenerated 127 logs for races out of 127 candidates |
oops.. the command was supposed to be "-c 200" no "-C 200". and: we used the latest verison of racepro as installed in hp05, (unclear how you got your results before...) On 08/04/2011 01:58 PM, Oren Laadan wrote:
|
Detail for the bug:
all: .obj .obj/abr2gbr
.obj:$@ ]; then mkdir $ @; fi
if [ ! -d
.obj/%.o: %.c
$(CC) -o $ @ $(CFLAGS) -c $ <
.obj/abr2gbr: .obj/abr2gbr.o
$(CC) -o $ @ $(CFLAGS) $ (LIBS) .obj/abr2gbr.o
So, mkdir() will race with open() system call made by gcc.
INFO:root: running post-replay callback...
INFO:root: time: 0.61
INFO:root: running pre-replay callback...
INFO:root: replaying ...
if [ ! -d .obj ]; then mkdir .obj; fi
INFO:root:
Backtrace:
[08] mkdir() = 0
[08] regs: eip: 0073:4001e424, eflags: 00000246, eax: 00000027, ebx: bf883daa, ecx: 000001ff, edx: bf883daa esi: bf881bb8 , edi: 00000001, ebp: bf881b48, esp: bf881a6c ds: 007b, es: 007b, fs: 0000, gs: 0033, ss: 007b
[08] data: input string, ptr = 0xbf883daa, size = 4, ".obj"
[08] resource lock, type = inode, access = write, id = 71, serial = 17, desc = none
[08] resource unlock, id = 71
Replay Diverged:
[05] diverged on event type = scribe_event_resource_lock_extra
RACE 132: replay diverge (200)
The text was updated successfully, but these errors were encountered: