Re: Memory checkers
áéåí ùðé, 9 áñôèîáø 2002, 19:36, voguemaster ëúá: There are plenty of open source utilities: memory patrol, electric fence, checker - just to name a few. I don't have time to check every one of them. What you can propose ? I've had good experience with DMALLOC and a bit of electric fence. I haven't been able to find a VERY GOOD tool for multithreaded appications :-\ Can you elaborate a bit more on your problem ? Are you absolutely sure it's related to memory leaks ? could it be something else (such as heap corruption) ?? Have you tried wrapping around malloc()/free() calls and print some debug information ? After the program stops you might parse the info and look for indescrepencies just a thought.. In the open source project I'm working on - Kannel, we have a malloc()/free() wrapper which reports memory leaks and keeps track of the function who owns each memory block. it's not fool proof, but at times have proven valuable at tracking down memory problems. -- Oded ::.. WWW is the MS-DOS of hypertext systems... -- Erik Naggum To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Memory checkers
Hello. Application I'm developing runs fine and fails only after two,three days of work under heavy load. I tried to use Insure++ from Parasoft but it adds a huge overhead - original application almost does not work. What other memory checkers can be used ? They have to have following features: 1. Small overhead 2. Support multithreading applications 3. Support C++ - unmangling function names. 4. Be more or less stable - not alpha/beta versions. 5. Have to know how to check one thing (for example memory leaks) but do it well. There are plenty of open source utilities: memory patrol, electric fence, checker - just to name a few. I don't have time to check every one of them. What you can propose ? Thanks, Michael. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Memory checkers
On Mon, Sep 09, 2002 at 01:27:28PM +0300, Michael Sternberg wrote: Hello. Application I'm developing runs fine and fails only after two,three days of work under heavy load. I tried to use Insure++ from Parasoft but it adds a huge overhead - original application almost does not work. What other memory checkers can be used ? valgrind. Best thing since sliced bread. -- Muli Ben-Yehuda syscalltrack hacker-at-large msg21640/pgp0.pgp Description: PGP signature
Re: Memory checkers
On Mon, 2002-09-09 at 13:27, Michael Sternberg wrote: Hello. Application I'm developing runs fine and fails only after two,three days of work under heavy load. I tried to use Insure++ from Parasoft but it adds a huge overhead - original application almost does not work. Interesting. In a previous job we tried to maek Insure++ to work with multithreaded applciations for several months(!) and although we had no problem running out applciation with it we didn't get from it almost any meaningfull output - it didn't find anything. Including problem we discovered with other means (staring at the source code) and we knew were there. Gilad. -- Gilad Ben-Yossef [EMAIL PROTECTED] http://benyossef.com We don't need kernel hackers or geniuses, we need good developers who will do what they're told. Famous last words, the collection. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Memory checkers
Valgrind may be problematic regarding threads. http://developer.kde.org/~sewardj/docs/manual.html#limits On Mon, 9 Sep 2002, Muli Ben-Yehuda wrote: On Mon, Sep 09, 2002 at 01:27:28PM +0300, Michael Sternberg wrote: Hello. Application I'm developing runs fine and fails only after two,three days of work under heavy load. I tried to use Insure++ from Parasoft but it adds a huge overhead - original application almost does not work. What other memory checkers can be used ? valgrind. Best thing since sliced bread. -- Orna. | http://tx.technion.ac.il/~agmon one penguin, two penguin, three penguin... = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Memory checkers
http://www.linuxjournal.com/article.php?sid=6059 On Mon, 9 Sep 2002 15:07:59 +0300 (IDT) Orna Agmon [EMAIL PROTECTED] wrote: Valgrind may be problematic regarding threads. http://developer.kde.org/~sewardj/docs/manual.html#limits On Mon, 9 Sep 2002, Muli Ben-Yehuda wrote: On Mon, Sep 09, 2002 at 01:27:28PM +0300, Michael Sternberg wrote: Hello. Application I'm developing runs fine and fails only after two,three days of work under heavy load. I tried to use Insure++ from Parasoft but it adds a huge overhead - original application almost does not work. What other memory checkers can be used ? valgrind. Best thing since sliced bread. -- Orna. | http://tx.technion.ac.il/~agmon one penguin, two penguin, three penguin... = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Memory checkers
Parasoft sell two Insure++ versions. The version which is supposed to be multi-thread compatible used to cost 500 extra dollars (3500 instead of 3000) for the linux version. could it be that you were using the wrong version? btw, now that the israeli sales person, moti berger, has vanished, the USA sales department are asking for even a higher price. On 9 Sep 2002, Gilad Ben-Yossef wrote: On Mon, 2002-09-09 at 13:27, Michael Sternberg wrote: Hello. Application I'm developing runs fine and fails only after two,three days of work under heavy load. I tried to use Insure++ from Parasoft but it adds a huge overhead - original application almost does not work. Interesting. In a previous job we tried to maek Insure++ to work with multithreaded applciations for several months(!) and although we had no problem running out applciation with it we didn't get from it almost any meaningfull output - it didn't find anything. Including problem we discovered with other means (staring at the source code) and we knew were there. Gilad. -- Orna. | http://tx.technion.ac.il/~agmon one penguin, two penguin, three penguin... = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Memory checkers
On Mon, 2002-09-09 at 17:23, Orna Agmon wrote: Parasoft sell two Insure++ versions. The version which is supposed to be multi-thread compatible used to cost 500 extra dollars (3500 instead of 3000) for the linux version. could it be that you were using the wrong version? No, it was the multi threaded version complete with the higher price tag :-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] http://benyossef.com We don't need kernel hackers or geniuses, we need good developers who will do what they're told. Famous last words, the collection. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Memory checkers
Gilad Ben-Yossef [EMAIL PROTECTED] writes: No, it was the multi threaded version complete with the higher price tag :-) Did you ask Parasoft for a refund? ;-) -- Oleg Goldshmidt | [EMAIL PROTECTED] = ... Of theoretical physics and programming, programming embodied the greater intellectual challenge. [E.W.Dijkstra, 1930 - 2002.] = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Memory checkers
There are plenty of open source utilities: memory patrol, electric fence, checker - just to name a few. I don't have time to check every one of them. What you can propose ? I've had good experience with DMALLOC and a bit of electric fence. I haven't been able to find a VERY GOOD tool for multithreaded appications :-\ Can you elaborate a bit more on your problem ? Are you absolutely sure it's related to memory leaks ? could it be something else (such as heap corruption) ?? Have you tried wrapping around malloc()/free() calls and print some debug information ? After the program stops you might parse the info and look for indescrepencies just a thought.. Eli = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]