On 5/11/09, Andreas Robinson <andr345@xxxxxxxxx> wrote: > On Mon, 2009-05-11 at 09:22 +0100, Alan Jenkins wrote: >> Andreas Robinson wrote: >> > >> > That the strings are malloc'd elsewhere should not be a problem if the >> > caller is aware that the strings are in the ELF-file and that they >> > remain available until release_file() is called. Depmod can't release >> > any of the modules until it is finished. >> > >> >> Withdrawn. >> >> Sorry, I'm too quick to jump to conclusions :-(. I thought the symbol >> names were being copied onto the heap, but they're clearly not. I'm not >> worried about string constants per se. I was mistakenly concerned that >> one data structure was being used to return strings with very different >> lifecycle rules. >> >> It's an important issue, so you might make this more explicit. One way >> would be to add an inline strtbl_free() (which collapses to free()). >> But you are also welcome to dismiss me as a hopeless reader and leave it >> as is:-). > > No you are not a hopeless reader. :) And I appreciate the feedback. I do > need to explain myself better. I never come up with a good solution on > the first attempt and usually need a few hours (or days) to figure out > the little details that make a particular design choice "obvious", even > though it's anything but. > >> > Oh, and could you please send me add --valgrind patch please? >> > >> > I wasn't subscribed to this list when you posted it and the archive has >> > mangled it somehow. (Or PEBKAC, more likely.) >> > >> >> Oh, sorry! I assumed it had already been applied. I'll resend it and >> see if it wakes Jon up. It might have been my fault for forgetting to >> put [PATCH] in the subject line. >> >> Btw, I'm being serious when I talk about it being slow. "valgrind >> ./depmod" should be enough to reproduce this one issue, so do that first >> :-). > > Ok, I will. :) A plain "time depmod" runs about 25 times faster than > "time valgrind depmod". The full test suite would then be about 10 - 12 > minutes on my machine ... Oops, sorry for wasting your time. If I'd looked more closely at the call trace I'd have realised this probably wasn't your fault. In fact, I think it's mine! Fortunately it doesn't affect normal usage. I'll fix this little embarrassment myself. Thanks Alan -- To unsubscribe from this list: send the line "unsubscribe linux-modules" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
