This is the mail archive of the binutils@sourceware.org mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: FAIL: Link with zlib-gabi compressed debug output


>>> On 07.12.15 at 15:59, <hjl.tools@gmail.com> wrote:
> On Mon, Dec 7, 2015 at 6:45 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>>> On 07.12.15 at 15:23, <hjl.tools@gmail.com> wrote:
>>> On Mon, Dec 7, 2015 at 5:47 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>> Is it really appropriate/necessary for the test to expect 1-byte
>>>> alignment:
>>>>
>>>> #...
>>>>  +\[[0-9a-f]+\]: .*COMPRESSED
>>>>  +ZLIB, [0-9a-f]+, 1
>>>> #pass
>>>>
>>>> (in ld-elf/gabinormal.rt)? I'm seeing .debug_aranges as the first
>>>> debug info section, emitted with 8-byte alignment:
>>>>
>>>> ...
>>>>   [28] .comment.SUSE.OPTs
>>>>        PROGBITS        00000000 000781 000006 01   0   0  1
>>>>        [00000030]: MERGE, STRINGS
>>>>   [29] .debug_aranges
>>>>        PROGBITS        00000000 000ef8 000047 00   0   0  8
>>>>        [00000800]: COMPRESSED
>>>>        ZLIB, 00000080, 8
>>>>   [30] .debug_pubnames
>>>>        PROGBITS        00000000 000f3f 00007c 00   0   0  1
>>>>        [00000800]: COMPRESSED
>>>>        ZLIB, 00000094, 1
>>>> ...
>>>
>>> Where does 8-byte alignment come from?
>>
>> I guess the compiler/assembler emits it? I didn't look at the details of
>> how the test binaries get created. It simply seems acceptable to me
>> for .debug_aranges to have other than 1-byte alignment.
> 
> It could be a real bug in binutils.  We need to get the bottom of
> this.

The host's crt1.o has an 8-byte aligned .debug_aranges. No
other object file or static library involved in the linking has other
than 1-byte alignment here.

Jan


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]