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: xtensa PR65730

On Fri, Apr 10, 2015 at 9:44 PM,
<> wrote:
> On Fri, Apr 10, 2015 at 11:40 AM,
> <> wrote:
>> On Fri, Apr 10, 2015 at 11:37 AM, Max Filippov <> wrote:
>>> On Fri, Apr 10, 2015 at 9:29 PM,
>>> <> wrote:
>>> > On Fri, Apr 10, 2015 at 6:31 AM, Max Filippov <> wrote:
>>> >> OTOH calling helper function to do multiplication by a constant 8 looks
>>> >> rather stupid. I guess we're not going to have non-8-bit bytes on xtensa
>>> >> anytime soon, maybe this multiplication can be replaced with shift?
>>> >
>>> > I think the optimizer will convert it after this point, so it
>>> > shouldn't matter. I would need to verify that though.
>>> It doesn't, I see the call in the final assembly code, even with various
>>> optimization levels.
>> This is likely a second bug then. It should have been eliminated prior to this call.
>> What happens on a config _with_ a multiply instruction? Do you get a multiply instruction, or a shift?

-O0 gives mull, higher levels give shift.

> Or possibly just a single bug in the sense that the rest of this code
> expects the multiply to be resolved to something else at this point.

Then configuration w/o multiplication should call helper at -O0 and
use shift at higher optimization levels?

-- Max

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