This is the mail archive of the
mailing list for the glibc project.
Re: [PATCH][BZ #19329] Fix race between tls allocation at thread creation and dlopen
- From: Adhemerval Zanella <adhemerval dot zanella at linaro dot org>
- To: libc-alpha at sourceware dot org
- Date: Mon, 18 Jan 2016 16:00:14 -0200
- Subject: Re: [PATCH][BZ #19329] Fix race between tls allocation at thread creation and dlopen
- Authentication-results: sourceware.org; auth=none
- References: <568D5E11 dot 3010301 at arm dot com> <5693D908 dot 8090104 at arm dot com> <5695B7BF dot 7050000 at redhat dot com> <5697BA1F dot 6010801 at arm dot com>
On 14-01-2016 13:09, Szabolcs Nagy wrote:
> On 13/01/16 02:34, Carlos O'Donell wrote:
>> On 01/11/2016 11:32 AM, Szabolcs Nagy wrote:
>>> 2016-01-11 Szabolcs Nagy <email@example.com>
>>> [BZ #19329]
>>> * elf/dl-open.c (dl_open_worker): Write GL(dl_tls_generation)
>>> * elf/dl-tls.c (_dl_allocate_tls_init): Read GL(dl_tls_generation),
>>> GL(dl_tls_max_dtv_idx), slotinfo entries and listp->next atomically.
>>> (_dl_add_to_slotinfo): Write the slotinfo entries and listp->next
>> You are headed in the right direction. I like where this patch is going,
>> but don't have enough time to review this in detail yet.
> will there be time before 2.23?
I intend to send a message about the hard freeze today and the question is
if this fix is suitable to not show regression in the following mount and
if so if we could iron out the bugs. So are you confident about that?
I am asking you this because this synchronization issues are a nest of rats
sometimes and I also noted you found out multiple issues during the patch
sending and reviewing. Objections?
>> At first glance your patch lacks sufficient concurrency documentation to
>> be acceptable. You need to document which acquires the releases
>> synchronizes-with. Please grep for "CONCURRENCY NOTES" to see the level
>> of detail we need to maintain these kinds of changes.
> i wanted to avoid documenting all the mess in the dynamic linker,
> but i can improve the comments.
> i see the following bugs:
> 1) pthread_create is not synced with dlopen
> 2) pthread_create is not synced with dlclose
> 3) tls access is not synced with dlopen
> 4) tls access is not synced with dlclose
> 5) tls access can oom crash
> 6) tls access is not as-safe
> 7) dlopen holds a global lock during ctors
> i can fix 1) by adding some atomics (this is the current patch).
> for 2) that's not enough, because dlclose has to wait with the
> free(link_map) while pthread_create is initializing the tls.
So which would be a better approach regarding it?
> i guess 3) can be fixed similarly to 1) but i don't have a
> test case for that.
> the rest is harder to fix.
> is it ok to only fix 1) for 2.23?
> or should i try to fix 3) as well (races on the same globals)?
I would say we attack one problem at time, even when they are related.
How hard do you consider to fix 3.?