This is the mail archive of the 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: [LTO] Patch to implement -flto-single (-flto-test) switch in gcc driver

William Maddox wrote:
> This patch adds the missing '-flto-test' functionality to the 'gcc'
> driver for LTO.


> Diego strongly objected to the name '-flto-test' as used in the LTO driver
> specification and in an earlier version of this patch.  He suggested
> '-flto-single'
> as a more descriptive name, and I have followed his suggestion here.

OK; I have no strong feelings about it.  (The reason I suggested -test
was that this mode isn't useful for anything except testing.)

> I could not find a way to deal with the '-flto-single' option within the specs
> language implemented by 'gcc'.   I thus extended the specs language
> to permit a spec function to be invoked as a predicate within the %{} construct,
> for example:
>     %{?lto-single():-flto}

Is that because of the -f in the option name?  So, something like:


doesn't work?  If so, perhaps we should just drop the "-f" as that would
avoid creating this new infrastructure just for this case.

Nathan, I know you've been running the DejaGNU testsuite with LTO.  Have
you been using Bill's patch, or have you cobbled some other mechanism to
do that?


Mark Mitchell
(650) 331-3385 x713

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