This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC 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: PING^3: [PATCH] i386: Don't generate alias for function return thunk


On Thu, Mar 15, 2018 at 12:54 PM, Jan Hubicka <hubicka@ucw.cz> wrote:
>> On Thu, Mar 15, 2018 at 10:47 AM, Jan Hubicka <hubicka@ucw.cz> wrote:
>> >> On Thu, Mar 15, 2018 at 9:03 AM, Jan Hubicka <hubicka@ucw.cz> wrote:
>> >> >> > What is the reason for using different names for return and indirect thunks at first place?
>> >> >> >
>> >> >>
>> >> >> These 2 thunks are identical.  But one may want to provide an
>> >> >> alternate thunk only for
>> >> >> indirect branch and leaves return thunk alone.  You can't do that if
>> >> >> both have the same
>> >> >> name.
>> >> >
>> >> > Hmm, OK, what is the benefit to have two different thunks? It is just
>> >> > safety precaution so we could adjust one without adjusting the other in
>> >> > future?
>> >> >
>> >>
>> >> That is correct.
>> >
>> > Hmm, I guess the patch is OK. Things are slightly more flexible this way and
>> > duplicating thunk is not terribly expensive. One can always link with
>> > non-comdat+ alias.
>> >
>>
>> That is true.  OK to backport to GCC 7 after a few days?
> OK.  I suppose you are testing return thunks on some real environment, like GCC bootstrap :)

Yes.

-- 
H.J.


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