Re: [PATCH 3/3] elf: Implement force_first handling in _dl_sort_maps_dfs

[Florian Weimer <fweimer@redhat.com>]

* Adhemerval Zanella Netto:

> On 15/08/22 11:30, Florian Weimer via Libc-alpha wrote:
>> As documented in a comment _dl_close_worker, the skipping is actually
>> needed for correctness.  It also seems less surprising if the
>> just-opened object is always initialized last, even in the presence
>> of cycles.
>
> I think it is BZ#28937, isn't? Also could you extend the explanation as you
> did in the last comment, the initial phrase sounds confusing.

I have expanded the commit message.

> Maybe extend the comment to say that not _dl_sort_maps_dfs will move
> the main object to front, so where previous you have the maps input
> as:
>
>   maps[0].l_name=elf/tst-bz28937-dir/tst-bz28937-a.so
>   maps[1].l_name=[...]/elf/tst-bz28937-dir/tst-bz28937-a1.so
>   maps[2].l_name=[...]/elf/tst-bz28937-dir/tst-bz28937-a2.so
>   maps[3].l_name=./libc.so.6
>   maps[4].l_name=elf/ld.so
>
> It will not be properly sorted as:
>
>   maps[0].l_name=elf/tst-bz28937-dir/tst-bz28937-a.so
>   maps[1].l_name=[...]/elf/tst-bz28937-dir/tst-bz28937-a1.so
>   maps[2].l_name=[...]/elf/tst-bz28937-dir/tst-bz28937-a2.so
>   maps[3].l_name=./libc.so.6
>   maps[4].l_name=elf/ld.so
>
> Instead of wrongly:
>
>   maps[0].l_name=[...]/elf/tst-bz28937-dir/tst-bz28937-a1.so
>   maps[1].l_name=[...]/elf/tst-bz28937-dir/tst-bz28937-a2.so
>   maps[2].l_name=elf/tst-bz28937-dir/tst-bz28937-a.so
>   maps[3].l_name=./libc.so.6
>   maps[4].l_name=elf/ld.so

I think with just three elements, it's a bit misleading because a cycle
of this size is rotated correctly by the naive approach.

>> diff --git a/elf/dl-sort-maps.c b/elf/dl-sort-maps.c
>> index 5b550b1e94..cd2d9c93fc 100644
>> --- a/elf/dl-sort-maps.c
>> +++ b/elf/dl-sort-maps.c
>> @@ -182,8 +182,9 @@ dfs_traversal (struct link_map ***rpo, struct link_map *map,
>>  
>>  static void
>>  _dl_sort_maps_dfs (struct link_map **maps, unsigned int nmaps,
>> -		   bool force_first __attribute__ ((unused)), bool for_fini)
>> +		   bool force_first, bool for_fini)
>>  {
>> +  struct link_map *first_map = maps[0];
>
> Move this to where it is actually used.

We need to copy it before it's overwritten by the sort.

>> +  /* Skipping the first object at maps[0] is not valid in general,
>> +     since traversing along object dependency-links may "find" that
>> +     first object even when it is not included in the initial order
>> +     (e.g. a dlopen()'ed shared object can have circular dependencies
>> +     linked back to itself). In such a case, traversing N-1 objects
>
> Double space after period and think we do not reference symbol using '()'.

Fixed (although I just copied that part of the comment).

>> +     will create a N-object result, and raise problems.  Instead,
>> +     force the object back into first place after sorting.  */
>> +  if (force_first && maps[0] != first_map)
>> +    {
>> +      struct link_map *saved = maps[0];
>> +      maps[0] = first_map;
>> +      int i = 1;
>> +      while (true)
>> +	{
>> +	  assert (i < nmaps);
>> +	  struct link_map *current = maps[i];
>> +	  maps[i] = saved;
>> +	  if (current == first_map)
>> +	     break;
>> +	  saved = current;
>> +	  ++i;
>> +	}
>> +    }
>>  }
>
> It sounds reasonable to keep the main object as the initial map, although
> it would slow down a bit normal dlclose.  I think it would be possible
> to optimize the memory move with memmove here, although not sure if it is
> worth.

If we make the assert less precise, memmove actually simplifies the
code.  I've made the change.

>>  void
>> diff --git a/elf/dso-sort-tests-1.def b/elf/dso-sort-tests-1.def
>> index 5f7f18ef27..4bf9052db1 100644
>> --- a/elf/dso-sort-tests-1.def
>> +++ b/elf/dso-sort-tests-1.def
>> @@ -64,3 +64,10 @@ output: b>a>{}<a<b
>>  tst-bz15311: {+a;+e;+f;+g;+d;%d;-d;-g;-f;-e;-a};a->b->c->d;d=>[ba];c=>a;b=>e=>a;c=>f=>b;d=>g=>c
>>  output(glibc.rtld.dynamic_sort=1): {+a[d>c>b>a>];+e[e>];+f[f>];+g[g>];+d[];%d(b(e(a()))a()g(c(a()f(b(e(a()))))));-d[];-g[];-f[];-e[];-a[<a<c<d<g<f<b<e];}
>>  output(glibc.rtld.dynamic_sort=2): {+a[d>c>b>a>];+e[e>];+f[f>];+g[g>];+d[];%d(b(e(a()))a()g(c(a()f(b(e(a()))))));-d[];-g[];-f[];-e[];-a[<g<f<a<b<c<d<e];}
>> +
>> +# Test that even in the presence of dependency loops involving dlopen'ed
>> +# object, that object is initialized last (and not unloaded prematurely).
>> +# Final destructor order is indeterminate due to the cycle.
>> +tst-bz28937: {+a;+b;-b;+c;%c};a->a1;a->a2;a2->a;b->b1;c->a1;c=>a1
>
> So main program:
>
>   1. dlopen 'a' and 'b';
>   2. dclose 'b';
>   3. dlopen 'c';
>   4. dlsym 'c' and calls fn_a from 'c';
>
> And we have a circle dependency where a depends of a2 and a2 depends on 'a'.
>
> Do we need to add a test for multiple circles? For instance where you have
> either another disjointed circle ({+d};d->d2;d2->d) and/or another circle
> in same dependency chain (a1->b;b1->a)?

Could you add this as a follow-up patch?  It does not seem strictly
related (and I think we already have other tests for unloading cycles).

Thanks,
Florian