On Fri, Jul 01, 2011 at 06:05:07AM +0200, Stefan Tauner wrote:
compile tested only. btw... why dont we wrap malloc to automatically include these checks? there is probably a patch somewhere that adds totally awesome shutdown code in OOM cases, but afaics we just print a warning and exit right now... it is just stupid to do the checks in the real code then.
Signed-off-by: Stefan Tauner stefan.tauner@student.tuwien.ac.at
It's definately better than what we have now (no error checking at all), so:
Acked-by: Uwe Hermann uwe@hermann-uwe.de
A few notes for later though:
- We should call any shutdown function which may need to be called in the respective places (and any free()s which may be needed).
- Later we should not exit(1) upon the error, but rather return an error code instead (e.g. FL_ERR_MALLOC or similar), at least in all public "API" functions which are meant to be in the upcoming libflashrom.
- A small xmalloc() wrapper (or use another name) would indeed be nice, too, but once we really return error codes etc. it's not really useful anymore I think.
Uwe.
On Tue, 12 Jul 2011 18:40:20 +0200 Uwe Hermann uwe@hermann-uwe.de wrote:
On Fri, Jul 01, 2011 at 06:05:07AM +0200, Stefan Tauner wrote:
compile tested only. btw... why dont we wrap malloc to automatically include these checks? there is probably a patch somewhere that adds totally awesome shutdown code in OOM cases, but afaics we just print a warning and exit right now... it is just stupid to do the checks in the real code then.
Signed-off-by: Stefan Tauner stefan.tauner@student.tuwien.ac.at
It's definately better than what we have now (no error checking at all), so:
Acked-by: Uwe Hermann uwe@hermann-uwe.de
thanks, r1370
A few notes for later though:
- We should call any shutdown function which may need to be called in the respective places (and any free()s which may be needed).
that would probably be handled by int programmer_shutdown(void) in flashrom.c at the moment. it should either be named differently or wrapped with a more generic function though(?)
- Later we should not exit(1) upon the error, but rather return an error code instead (e.g. FL_ERR_MALLOC or similar), at least in all public "API" functions which are meant to be in the upcoming libflashrom.
and in the internal ones? just as errors values?
- A small xmalloc() wrapper (or use another name) would indeed be nice, too, but once we really return error codes etc. it's not really useful anymore I think.
that's true, at least as long as it is not a macro that automagically returns on errors (which would be super evil, but... well no "but" i guess. :)
-
Stefan Tauner -
Uwe Hermann