On Mon, 27 Jan 2025 at 13:30, Torbjorn SVENSSON
<torbjorn.svens...@foss.st.com> wrote:
>
> Hi Christophe,
>
> On 2025-01-27 13:07, Christophe Lyon wrote:
> > Hi Torbjorn,
> >
> > On Fri, 24 Jan 2025 at 18:45, Torbjorn SVENSSON
> > <torbjorn.svens...@foss.st.com> wrote:
> >>
> >> Another ping... :)
> >>
> >> Kind regards,
> >> Torbjörn
> >>
> >> On 2024-12-18 18:35, Torbjorn SVENSSON wrote:
> >>> Gentle ping :)
> >>>
> >>> Kind regards,
> >>> Torbjörn
> >>>
> >>> On 2024-11-14 17:16, Torbjorn SVENSSON wrote:
> >>>>
> >>>>
> >>>> On 2024-11-14 16:32, Christophe Lyon wrote:
> >>>>> On Fri, 8 Nov 2024 at 19:49, Torbjörn SVENSSON
> >>>>> <torbjorn.svens...@foss.st.com> wrote:
> >>>>>>
> >>>>>> Ok for trunk and releases/gcc-14?
> >>>>>>
> >>>>>> --
> >>>>>>
> >>>>>> A long time ago, this test forced -march=armv6.
> >>>>>>
> >>>>>> With -marm, the generated assembler is:
> >>>>>> foo:
> >>>>>> sub r0, r0, #48
> >>>>>> cmp r0, #9
> >>>>>> movhi r0, #0
> >>>>>> movls r0, #1
> >>>>>> bx lr
> >>>>>>
> >>>>>> With -mthumb, the generated assembler is:
> >>>>>> foo:
> >>>>>> subs r0, r0, #48
> >>>>>> movs r2, #9
> >>>>>> uxtb r3, r0
> >>>>>> movs r0, #0
> >>>>>> cmp r2, r3
> >>>>>> adcs r0, r0, r0
> >>>>>> uxtb r0, r0
> >>>>>> bx lr
> >>>>>>
> >>>>>> Require effective-target arm_arm_ok to skip the test for thumb-only
> >>>>>> targets (Cortex-M).
> >>>>>>
> >>>>>> gcc/testsuite/ChangeLog:
> >>>>>>
> >>>>>> * gcc.target/arm/unsigned-extend-1.c: Use effective-target
> >>>>>> arm_arm_ok.
> >>>>>>
> >>>>>> Signed-off-by: Torbjörn SVENSSON <torbjorn.svens...@foss.st.com>
> >>>>>> ---
> >>>>>> gcc/testsuite/gcc.target/arm/unsigned-extend-1.c | 1 +
> >>>>>> 1 file changed, 1 insertion(+)
> >>>>>>
> >>>>>> diff --git a/gcc/testsuite/gcc.target/arm/unsigned-extend-1.c b/gcc/
> >>>>>> testsuite/gcc.target/arm/unsigned-extend-1.c
> >>>>>> index 3b4ab048fb0..73f2e1a556d 100644
> >>>>>> --- a/gcc/testsuite/gcc.target/arm/unsigned-extend-1.c
> >>>>>> +++ b/gcc/testsuite/gcc.target/arm/unsigned-extend-1.c
> >>>>>> @@ -1,4 +1,5 @@
> >>>>>> /* { dg-do compile } */
> >>>>>> +/* { dg-require-effective-target arm_arm_ok } */
> >>>>>> /* { dg-options "-O2" } */
> >>>>>>
> >>>>>
> >>>>> So I'm guessing arm_arm_ok fails when you are testing for M-profile
> >>>>> targets?
> >>>>> What happens when you test for A-profile, providing -mthumb as part of
> >>>>> your runtest flags?
> >>>>
> >>>>
> >
> > Looking at our CI results, without your patch the test passes on:
> > cortex-m33, m3, m55, m7, v7A in arm mode, v8A in thumb mode,
> > and it fails only on cortex-m0 and m23.
> > Does this match your observations?
>
> Without my patch, I see failure on M0, M0+ and M23.
> With my patch, it goes to unsupported on all Cortex-M targets that I test.
>
> > So your patch will make the test unsupported on m33, m3, m55 and m7
> > while it currently passes?
> >
> > Am I missing something?
>
> I think you might have missed my comment down....
>
> >
> > Thanks,
> >
> > Christophe
> >
> >>>>
> >>>> For Cortex-M0 (thumb/cpu=cortex-m0/float-abi=soft):
> >>>>
> >>>> Testing arm/unsigned-extend-1.c
> >>>> check_compile tool: gcc for arm_arm_ok
> >>>> doing compile
> >>>> Executing on host: .../bin/arm-none-eabi-gcc arm_arm_ok21114.c -
> >>>> mthumb -mcpu=cortex-m0 -mfloat-abi=soft -fdiagnostics-plain-output
> >>>> -marm - Wno-complain-wrong-lang -S -o arm_arm_ok21114.s
> >>>> (timeout = 800)
> >>>> spawn -ignore SIGHUP .../bin/arm-none-eabi-gcc arm_arm_ok21114.c -
> >>>> mthumb -mcpu=cortex-m0 -mfloat-abi=soft -fdiagnostics-plain-output -
> >>>> marm -Wno- complain-wrong-lang -S -o arm_arm_ok21114.s
> >>>> pid is 21254 -21254
> >>>> cc1: error: target CPU does not support ARM mode
> >>>> pid is -1
> >>>> close result is 21254 exp6 0 1
> >>>> output is cc1: error: target CPU does not support ARM mode
> >>>> status 1
> >>>> compiler exited with status 1
> >>>> UNSUPPORTED: gcc.target/arm/unsigned-extend-1.c
> >>>>
> >>>>
> >>>> For Cortex-A7 (thumb/cpu=cortex-a7/float-abi=soft):
> >>>>
> >>>> Testing arm/unsigned-extend-1.c
> >>>> check_compile tool: gcc for arm_arm_ok
> >>>> doing compile
> >>>> Executing on host: .../bin/arm-none-eabi-gcc arm_arm_ok21296.c -
> >>>> mthumb -mcpu=cortex-a7 -mfloat-abi=soft -fdiagnostics-plain-output
> >>>> -marm - Wno-complain-wrong-lang -S -o arm_arm_ok21296.s
> >>>> (timeout = 800)
> >>>> spawn -ignore SIGHUP .../bin/arm-none-eabi-gcc arm_arm_ok21296.c -
> >>>> mthumb -mcpu=cortex-a7 -mfloat-abi=soft -fdiagnostics-plain-output -
> >>>> marm -Wno- complain-wrong-lang -S -o arm_arm_ok21296.s
> >>>> pid is 21434 -21434
> >>>> pid is -1
> >>>> output is status 0
> >>>> doing compile
> >>>> Executing on host: .../bin/arm-none-eabi-gcc .../gcc/testsuite/
> >>>> gcc.target/arm/unsigned-extend-1.c -mthumb -mcpu=cortex-a7 -mfloat-
> >>>> abi=soft -fdiagnostics-plain-output -O2 -ffat-lto-objects -fno-
> >>>> ident -S -o unsigned-extend-1.s (timeout = 800)
> >>>> spawn -ignore SIGHUP .../bin/arm-none-eabi-gcc .../gcc/testsuite/
> >>>> gcc.target/arm/unsigned-extend-1.c -mthumb -mcpu=cortex-a7 -mfloat-
> >>>> abi=soft -fdiagnostics-plain-output -O2 -ffat-lto-objects -fno-ident -
> >>>> S -o unsigned-extend-1.s
> >>>> pid is 21438 -21438
> >>>> pid is -1
> >>>> output is status 0
> >>>> check_compile tool: gcc for exceptions_enabled
> >>>> doing compile
> >>>> Executing on host: .../bin/arm-none-eabi-gcc
> >>>> exceptions_enabled21296.cc -mthumb -mcpu=cortex-a7 -mfloat-
> >>>> abi=soft -fdiagnostics-plain-output -Wno-complain-wrong-lang -S
> >>>> -o exceptions_enabled21296.s (timeout = 800)
> >>>> spawn -ignore SIGHUP .../bin/arm-none-eabi-gcc
> >>>> exceptions_enabled21296.cc -mthumb -mcpu=cortex-a7 -mfloat-abi=soft -
> >>>> fdiagnostics-plain-output -Wno-complain-wrong-lang -S -o
> >>>> exceptions_enabled21296.s
> >>>> pid is 21442 -21442
> >>>> pid is -1
> >>>> output is status 0
> >>>> PASS: gcc.target/arm/unsigned-extend-1.c (test for excess errors)
> >>>> PASS: gcc.target/arm/unsigned-extend-1.c scan-assembler-not uxtb
> >>>>
> >>>> For Cortex-A7 (cpu=cortex-a7/float-abi=soft):
> >>>>
> >>>>
> >>>>
> >>>> Testing arm/unsigned-extend-1.c
> >>>> check_compile tool: gcc for arm_arm_ok
> >>>> doing compile
> >>>> Executing on host: .../bin/arm-none-eabi-gcc arm_arm_ok21484.c -
> >>>> mcpu=cortex-a7 -mfloat-abi=soft -fdiagnostics-plain-output -marm -
> >>>> Wno-complain-wrong-lang -S -o arm_arm_ok21484.s (timeout = 800)
> >>>> spawn -ignore SIGHUP .../bin/arm-none-eabi-gcc arm_arm_ok21484.c -
> >>>> mcpu=cortex-a7 -mfloat-abi=soft -fdiagnostics-plain-output -marm -Wno-
> >>>> complain-wrong-lang -S -o arm_arm_ok21484.s
> >>>> pid is 21622 -21622
> >>>> pid is -1
> >>>> output is status 0
> >>>> doing compile
> >>>> Executing on host: .../bin/arm-none-eabi-gcc .../gcc/testsuite/
> >>>> gcc.target/arm/unsigned-extend-1.c -mcpu=cortex-a7 -mfloat-abi=soft
> >>>> - fdiagnostics-plain-output -O2 -ffat-lto-objects -fno-ident -S -
> >>>> o unsigned-extend-1.s (timeout = 800)
> >>>> spawn -ignore SIGHUP .../bin/arm-none-eabi-gcc .../gcc/testsuite/
> >>>> gcc.target/arm/unsigned-extend-1.c -mcpu=cortex-a7 -mfloat-abi=soft -
> >>>> fdiagnostics-plain-output -O2 -ffat-lto-objects -fno-ident -S -o
> >>>> unsigned-extend-1.s
> >>>> pid is 21626 -21626
> >>>> pid is -1
> >>>> output is status 0
> >>>> check_compile tool: gcc for exceptions_enabled
> >>>> doing compile
> >>>> Executing on host: .../bin/arm-none-eabi-gcc
> >>>> exceptions_enabled21484.cc -mcpu=cortex-a7 -mfloat-abi=soft -
> >>>> fdiagnostics-plain-output -Wno- complain-wrong-lang -S -o
> >>>> exceptions_enabled21484.s (timeout = 800)
> >>>> spawn -ignore SIGHUP .../bin/arm-none-eabi-gcc
> >>>> exceptions_enabled21484.cc -mcpu=cortex-a7 -mfloat-abi=soft -
> >>>> fdiagnostics-plain-output -Wno-complain-wrong-lang -S -o
> >>>> exceptions_enabled21484.s
> >>>> pid is 21630 -21630
> >>>> pid is -1
> >>>> output is status 0
> >>>> PASS: gcc.target/arm/unsigned-extend-1.c (test for excess errors)
> >>>> PASS: gcc.target/arm/unsigned-extend-1.c scan-assembler-not uxtb
> >>>> testcase .../gcc/testsuite/gcc.target/arm/arm.exp completed in 1 seconds
> >>>>
> >>>>> Since you don't add -marm, the test will be compiled in Thumb mode,
> >>>>> and still fail?
> >>>>
> >>>> Since my tests succeeded without it, it didn't think much of it. I can
> >>>> add -marm to the do-options just be safe. Should I send a V2 with it
> >>>> or can it be approved with that addition?
>
> ... here.
> While arm_arm_ok is checking that it's possible to use -marm, I suppose
> we should also add the -marm to the dg-additional-options in the test.
>
No, I had read this comment :-)
> Is this enough to have a pass on your CI?
> Would it be easier for you if I send a V2 with that change?
I don't expect any impact of the CI side, since the test passes in
thumb mode on A-profile anyway. But yes, that would be more
consistent....
My question (probably for Richard) is rather: is it OK to lose the
PASSes on m3/m7/m33/m55?
>
> Based on the history, I assumed that it was expected for this test case
> to be executed in arm mode. I suppose we could do it in thumb2 mode
> instead or something similar, but not sure if that's better...?
I'm not sure: the commit messages for this file are not very informative...
Just realized that
/* { dg-skip-if "" { arm_thumb1 } } */
might do what you want?
Thanks,
Christophe
>
> Kind regards,
> Torbjörn
>
>
> >>>>
> >>>> Kind regards,
> >>>> Torbjörn
> >>>>
> >>>>>
> >>>>> Thanks,
> >>>>>
> >>>>> Christophe
> >>>>>
> >>>>>> unsigned char foo (unsigned char c)
> >>>>>> --
> >>>>>> 2.25.1
> >>>>>>
> >>>>
> >>>
> >>
>
