tech-toolchain archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
Re: writing a (userland) core dump with incomplete info
In article <20241228135149.GB22347%mail.duskware.de@localhost>,
Martin Husemann <martin%duskware.de@localhost> wrote:
>Hey folks,
>
>a mabye stupid question: PR 58936 describes a userland process smashing
>its memory and triggering a core dump, but due to the memory corruption
>we get invalid addresses for the environmen and then fail to find the
>proper auxv vector:
>
>proc_getauxv(pid 21737 cmd swipl) -
> ps_envstr 0x697220616e737765 nenvstr 1920154122 auxv 0x69722065020d47bd
>
>
>proc_getauxv() then fails with EFAULT and we fail writing the auxv note
>in the core dump, and then stop the whole core writing process early, leaving
>the user with a zero sized swipl.core file.
>
>Can we do better? Does it make sense try debugging stuff w/o the auxv vector?
Well, perhaps the core writer can be more careful about making sure that
the environment is well-formed.
christos
Home |
Main Index |
Thread Index |
Old Index