sysvsem module: release_sysvsem_sem and lockup(semaphore leak)

php.internals

Yasuo Ohgaki

21 years ago
Since release_sysvsem_sem has error message while releasing acquired semaphores (both PHP_4_4 and PHP_5_1), it may end up with semaphore leak when error handler is used. The error message is useful for debugging, so the error message output should be configurable with ini setting. Although this workaround is not perfect solution, but it works in most cases. If there isn't a better idea, I'll add new ini directive to the module.
-- Yasuo Ohgaki

Derick Rethans

21 years ago
On Mon, 27 Jun 2005, Yasuo Ohgaki wrote:
> Since release_sysvsem_sem has error message while releasing acquired > semaphores (both PHP_4_4 and PHP_5_1), it may end up with semaphore > leak when error handler is used. > > The error message is useful for debugging, so the error message > output should be configurable with ini setting. Although this workaround > is not perfect solution, but it works in most cases. If there isn't a better > idea, I'll add new ini directive to the module.
We should not add more ini settings, we've enough of them already. Derick
-- Derick Rethans http://derickrethans.nl | http://ez.no | http://xdebug.org

Yasuo Ohgaki

21 years ago
Derick Rethans wrote:
> On Mon, 27 Jun 2005, Yasuo Ohgaki wrote: > > >>Since release_sysvsem_sem has error message while releasing acquired >>semaphores (both PHP_4_4 and PHP_5_1), it may end up with semaphore >>leak when error handler is used. >> >>The error message is useful for debugging, so the error message >>output should be configurable with ini setting. Although this workaround >>is not perfect solution, but it works in most cases. If there isn't a better >>idea, I'll add new ini directive to the module. > > > We should not add more ini settings, we've enough of them already. > > Derick >
I'll just comment out problematic line that may cause server lockup.
-- Yasuo Ohgaki

Andi Gutmans

21 years ago
Yasuo, Isn't it possible to resolve this in the code directly? At 09:11 AM 6/27/2005 +0200, Derick Rethans wrote: