|
From: parag s. <par...@gm...> - 2005-07-20 17:17:35
|
hey i am having the following problem....=20 i want to run valdring on a process that does not die by itself. the only= =20 way to kill it is usually kill or cltr c . however i cant do that without= =20 killing valgrind.=20 so i was wondering if valgrind creates some log file of some kind which it= =20 keeps on updating all the time that the process is running and then uses=20 this log file at the end of the process to calculate the memory leaks.=20 if this is present then i can just kill valgrind and then ask it to analyse= =20 that log file to give me the leak summary.=20 i would really appretiate it if the developers can let me know if something= =20 of this sort is possible.=20 thanks parag shah |
|
From: Brian C. <cr...@fi...> - 2005-07-20 17:24:03
|
You can send other signals to your process using kill... if interrupts and sigquit are being eaten by valgrind, try using one of the SIGUSR signals. -- Brian parag shah wrote: > hey > > i am having the following problem.... > > i want to run valdring on a process that does not die by itself. the > only way to kill it is usually kill or cltr c . however i cant do that > without killing valgrind. > > so i was wondering if valgrind creates some log file of some kind which > it keeps on updating all the time that the process is running and then > uses this log file at the end of the process to calculate the memory leaks. > > if this is present then i can just kill valgrind and then ask it to > analyse that log file to give me the leak summary. > > i would really appretiate it if the developers can let me know if > something of this sort is possible. > > > thanks > parag shah > |
|
From: Nicholas N. <nj...@cs...> - 2005-07-20 18:26:09
|
On Wed, 20 Jul 2005, parag shah wrote: > i want to run valdring on a process that does not die by itself. the only > way to kill it is usually kill or cltr c . however i cant do that without > killing valgrind. > > so i was wondering if valgrind creates some log file of some kind which it > keeps on updating all the time that the process is running and then uses > this log file at the end of the process to calculate the memory leaks. > > if this is present then i can just kill valgrind and then ask it to analyse > that log file to give me the leak summary. If you Ctrl-C a program running under Valgrind it should terminate and then Valgrind should do it's normal termination actions, including leak checking when you are using Memcheck. So maybe something else is going wrong. You'll have to give more details, such as the program you are running and how you are invoking, what version of Valgrind, etc. If you can produce a small test case that would help a lot. N |
|
From: Nicholas N. <nj...@cs...> - 2005-07-21 00:15:16
|
On Wed, 20 Jul 2005, parag shah wrote: > ya this works. what i was actually doing is running valgrind as a backgrond > process and thus could not do a cltr c. doing a kill -9 didnt work. > > if i dont run it with & then i can do a cltr c which propogates to my > precess, kills it and then valgrind can calculate the memory leak. kill -9 sends "SIGKILL" which no process can catch, which is why Valgrind doesn't do the leak checking. Use "kill -2" or "kill -15" (SIGTERM and SIGINT respectively) instead and Valgrind will do leak checking. N |