Re: [perl #34501] Segfault in stress test
Leopold Toetsch
> I've been checking that my md5sum library still works after a few
> months,
Yeah. PIR code is quite stable.
> ... but it seems to segfault in a test file with the
> following backtrace:
> #0 0x0816d0d0 in do_thaw (interpreter=0x82f7050, pmc=0x84be710,
> info=0xbfffef90) at src/pmc_freeze.c:1017
> (gdb) print *info->thaw_ptr
> Cannot access memory at address 0x41143c1c
> The segfault occurs in one of the $P0 = _config() calls, the one in
> _md5_print.
Yet another GC bug. Seems that the hash got collected during thawing it.
I've disabled DOD runs inside thaw now.
Thanks for reporting,
leo
Nick Glencross
>Nick Glencross <parrotbug...@parrotcode.org> wrote:
>
>
>
>>I've been checking that my md5sum library still works after a few
>>months,
>>
>>
>
>Yeah. PIR code is quite stable.
>
>
It is, isn't it? No changes were need what-so-ever.
> Thanks for reporting,
>
>leo
>
>
I do hope so. I rarely know what the 'right solution' is, which is why
leaving that choice to you is often best...
Cheers,
Nick
Nick Glencross
>Nick Glencross <parrotbug...@parrotcode.org> wrote:
>
>
>>I've been checking that my md5sum library still works after a few
>>months,
>>
>>
>
>I've now checked in the whole MD5 stuff, slightly modifed (whitespace,
>cosmetics, converted to a library, fix error with files > 2MB) and a bit
>faster (25 times for bigger files :)
>
>Thanks for the contribution,
>leo
>
My hero. Thanks! Has anyone had a chance to try it on 64 bit, or swapped
endian?
Nick