This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: vi and glibc 2.12: Assertion `_rtld_global_ro._dl_pagesize != 0
On Wed, Aug 18, 2010 at 8:47 AM, Carlos O'Donell wrote:
> On Wed, Aug 18, 2010 at 1:31 AM, Mike Frysinger wrote:
>> On Tue, Aug 17, 2010 at 10:36 PM, Mike Frysinger wrote:
>>> i'm certainly not familiar with glibc's ldso implementation due to
>>> it's crazy large & complicated code base, but i imagine this could be
>>> resolved best by having the ldso declare an initializer that calls the
>>> same functions as the ldso does when bootstrapping a normal
>>> application (_dl_sysdep_start/etc...). ?this is because the nss libs
>>> are dlopened like any other ELF, and it tails into libc.so which tails
>>> into the ldso. ?there is no special code that i can see that goes
>>> "this is the ldso, do xxx specially" which makes perfect sense. ?only
>>> outstanding issue there would be whether the ldso will process its own
>>> initializers when bootstrapping a standard application.
>>>
>>> we could back out the assert in __getpagesize(), but that would only
>>> "fix" things until the next time someone adds code which relies on
>>> initialized GLRO() state.
>>
>> after some experimenting, i guess the init method isnt feasible. ?as
>> the GLRO() name indicates, the struct is backed by read-only mappings,
>> and the ELF processing has already set up all the mappings with proper
>> permissions per the program headers by the time it calls any
>> initializers for that ELF. ?temporarily remapping itself isnt feasible
>> as some kernels prevent re-enabling write access to a read/execute
>> only mapping.
>
> Does DL_STATIC_INIT solve the problem?
>
> See dl-static.c as implemented by ia64 and mips.
i stumbled across this file earlier, but didnt quite fathom what it
was doing. now that you point it out in light of my other info, it
makes perfect sense.
looking at the ia64 code, this is the exact reason it is doing
DL_STATIC_INIT in the first place ... it's been assuming dl_pagesize
forever. now that common linux supports that, we can do the same
there. seems to work for me ... i'll post a patch once i finish
building things locally.
-mike