This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [PATCH] gas: fix a few omissions in .cfi_label handling
- From: "Jan Beulich" <JBeulich at suse dot com>
- To: "H.J. Lu" <hjl dot tools at gmail dot com>
- Cc: "Binutils" <binutils at sourceware dot org>
- Date: Thu, 05 Feb 2015 16:28:28 +0000
- Subject: Re: [PATCH] gas: fix a few omissions in .cfi_label handling
- Authentication-results: sourceware.org; auth=none
- References: <54D39521020000780005D651 at mail dot emea dot novell dot com> <CAMe9rOrMnQ4yAPqbPq55uPOJRrAd2Uz4nmo-g34pDU11o3n9Nw at mail dot gmail dot com> <54D39E1A020000780005D6C4 at mail dot emea dot novell dot com> <CAMe9rOqhZRhiOjvnaCKjDD2v5yYnM2n5a-nt-PHPUVVz-rb_nw at mail dot gmail dot com> <54D3A5BE020000780005D741 at mail dot emea dot novell dot com> <CAMe9rOodO-GQ071TWG1iWxcaWsrEO623-rTWVrVSOxXp668EUQ at mail dot gmail dot com>
>>> On 05.02.15 at 17:21, <hjl.tools@gmail.com> wrote:
> On Thu, Feb 5, 2015 at 8:17 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>>> On 05.02.15 at 16:50, <hjl.tools@gmail.com> wrote:
>>> On Thu, Feb 5, 2015 at 7:45 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>>>>> On 05.02.15 at 16:38, <hjl.tools@gmail.com> wrote:
>>>>> On Thu, Feb 5, 2015 at 7:06 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>>>> While actually starting to use that new directive, I noticed a few
>>>>>> oversights of the original commit.
>>>>>>
>>>>>> gas/
>>>>>> 2015-02-05 Jan Beulich <jbeulich@suse.com>
>>>>>>
>>>>>> * dw2gencfi.c (select_cie_for_fde): Also bail on CFI_label.
>>>>>> (cfi_change_reg_numbers): Also do nothing for CFI_label.
>>>>>> (cfi_pseudo_table): Also handle .cfi_label when not supporting
>>>>>> CFI directives.
>>>>>>
>>>>>
>>>>> Please add some testcases.
>>>>
>>>> For what? The original patch added some already, and I really don't
>>>> see the point of me wasting time on trying to figure out how I could
>>>> demonstrate that things could go wrong without these (minor and
>>>> obvious) adjustments.
>>>
>>> How does one know what the new directive should
>>> behave? Without the testcase, it may be broken and
>>> we won't even know it.
>>
>> There is no new directive being added here. All that happens are a
>> couple of adjustments to the code that added it recently (with - upon
>> your request - a test case).
>>
>
> Doesn't your new patch add cfi_label? If it is just a dummy,
> why not just replace those dummy CFI directives with a
> cfi_dummy directive?
It adds (fake) support for the directive in the case when CFI
directives aren't supported (as said in the last log entry still
quoted above). There are two tables, and it's only the latter
(dummy) one that I forgot to update in the original patch.
Jan