Messages in this thread | | | Date | Mon, 29 Jun 2009 13:45:53 +0300 | From | Sergey Senozhatsky <> | Subject | Re: kmemleak hexdump proposal |
| |
On (06/29/09 13:19), Pekka Enberg wrote: > Hi Sergey, > > On Mon, 2009-06-29 at 13:19 +0300, Sergey Senozhatsky wrote: > > Well, it's not easy to come up with something strong. > > I agree, that stack gives you almost all you need. > > > > HEX dump can give you a _tip_ in case you're not sure. > > Don't get me wrong, I'm not against it in any way. I don't think so ;) _IF_ even so - it's absolutely normal, to my mind.
> If Catalin is interested in merging this kind of > functionality, go for it! You might want to consider unconditionally > enabling the hexdump. If the information is valuable, we should print > it all the time. I guess it's valuable enougth to print it, but not valuable enougth to print it all the time. So, let me say - it's valuable enougth to print 'on demand', I guess. (I may be wrong).
BTW, printing it all the time we can spam kmemleak (in case there are objects sized 2K, 4K and so on). That's why I wrote about hexdump=OBJECT_POINTER.
> > Pekka >
Sergey
| |