svn commit: r255983 - head/libexec/rtld-elf

[Ed Maste]

On 2 October 2013 01:29, Konstantin Belousov <kostikbel at gmail.com> wrote:
> On Wed, Oct 02, 2013 at 02:32:59AM +0000, Ed Maste wrote:
>> Modified: head/libexec/rtld-elf/rtld.c
>> ==============================================================================
>> --- head/libexec/rtld-elf/rtld.c      Wed Oct  2 00:50:27 2013        (r255982)
>> +++ head/libexec/rtld-elf/rtld.c      Wed Oct  2 02:32:58 2013        (r255983)
>> @@ -1111,11 +1111,7 @@ digest_dynamic1(Obj_Entry *obj, int earl
>>               break;
>>
>>       case DT_MIPS_RLD_MAP:
>> -#ifdef notyet
>> -             if (!early)
>> -                     dbg("Filling in DT_DEBUG entry");
>> -             ((Elf_Dyn*)dynp)->d_un.d_ptr = (Elf_Addr) &r_debug;
>> -#endif
>> +             *((Elf_Addr *)(dynp->d_un.d_ptr)) = (Elf_Addr) &r_debug;
>>               break;
>>  #endif
>>
> Could the DT_DEBUG and DT_MIPS_RLD_MAP cases be collapsed some more, e.g. by
> putting only the 'case <something>:' lines under ifdef ?

No, the case statement bodies are different as well:

DT_DEBUG:
((Elf_Dyn*)dynp)->d_un.d_ptr = (Elf_Addr) &r_debug;

DT_MIPS_RLD_MAP:
*((Elf_Addr *)(dynp->d_un.d_ptr)) = (Elf_Addr) &r_debug;

> Also, you removed the if (!early) test.

It was just a dbg() statement in the if (!early) case; the assignment
was unconditional (well, aside from being #ifdef'd away).