[PATCH] ld: xfail riscv64be-*-* for ld-scripts/empty-address-2 tests.

[PATCH] ld: xfail riscv64be-*-* for ld-scripts/empty-address-2 tests.

[Nelson Chu]

From: Marcus Comstedt <marcus@mc.pp.se>

Same as the the riscv64-*-* (little endian), we need to xfail
riscv64be-*-* (big endian), too.

ld/
    * testsuite/ld-scripts/empty-address-2a.d: xfail riscv64-*-*.
    * testsuite/ld-scripts/empty-address-2b.d: Likewise.
---
 ld/testsuite/ld-scripts/empty-address-2a.d | 2 +-
 ld/testsuite/ld-scripts/empty-address-2b.d | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/ld/testsuite/ld-scripts/empty-address-2a.d b/ld/testsuite/ld-scripts/empty-address-2a.d
index 3b5a2d6..2619c48 100644
--- a/ld/testsuite/ld-scripts/empty-address-2a.d
+++ b/ld/testsuite/ld-scripts/empty-address-2a.d
@@ -1,7 +1,7 @@
 #source: empty-address-2.s
 #ld: -Ttext 0x0000000 -Tdata 0x200 -T empty-address-2a.t
 #nm: -n
-#xfail: frv-*-*linux* riscv64-*-* tic54x-*-*
+#xfail: frv-*-*linux* riscv64-*-* riscv64be-*-* tic54x-*-*
 #...
 0+0 T _start
 #...
diff --git a/ld/testsuite/ld-scripts/empty-address-2b.d b/ld/testsuite/ld-scripts/empty-address-2b.d
index 3b530c9..ebe8461 100644
--- a/ld/testsuite/ld-scripts/empty-address-2b.d
+++ b/ld/testsuite/ld-scripts/empty-address-2b.d
@@ -1,7 +1,7 @@
 #source: empty-address-2.s
 #ld: -Ttext 0x0000000 -Tdata 0x200 -T empty-address-2b.t
 #nm: -n
-#xfail: frv-*-*linux* riscv64-*-* tic54x-*-*
+#xfail: frv-*-*linux* riscv64-*-* riscv64be-*-* tic54x-*-*
 #...
 0+0 T _start
 #...
-- 
2.7.4

Re: [PATCH] ld: xfail riscv64be-*-* for ld-scripts/empty-address-2 tests.

[Alan Modra]

On Wed, Jan 06, 2021 at 02:36:54AM -0800, Nelson Chu wrote:
> -#xfail: frv-*-*linux* riscv64-*-* tic54x-*-*
> +#xfail: frv-*-*linux* riscv64-*-* riscv64be-*-* tic54x-*-*

Not that it matters very much, but you could write riscv64*-*-*
instead of the two riscv entries.

-- 
Alan Modra
Australia Development Lab, IBM

Re: [PATCH] ld: xfail riscv64be-*-* for ld-scripts/empty-address-2 tests.

[Nelson Chu]

Hi Alan,

On Thu, Jan 7, 2021 at 8:06 AM Alan Modra <amodra@gmail.com> wrote:
>
> On Wed, Jan 06, 2021 at 02:36:54AM -0800, Nelson Chu wrote:
> > -#xfail: frv-*-*linux* riscv64-*-* tic54x-*-*
> > +#xfail: frv-*-*linux* riscv64-*-* riscv64be-*-* tic54x-*-*
>
> Not that it matters very much, but you could write riscv64*-*-*
> instead of the two riscv entries.

Thanks, these look better, I have updated them and committed.


Besides, riscv[32|64]be also get the following xpass,

XPASS: ld-elf/compressed1d

I think the RISC-V big-endian support won't break the compress rules,
just break the compressed1d test by accident.  Since the sizes of
compressed and uncompressed .debug_aranges are the same for the
riscv[32|64]be, but different for the little endian targets...  Jim
already had written an alternative test compressed1d-alt.  Therefore,
should we need to only xfail the riscv "little endian" targets for
ld-elf/compressed1d, but allow the big endian passes?

Thanks
Nelson

Re: [PATCH] ld: xfail riscv64be-*-* for ld-scripts/empty-address-2 tests.

[Alan Modra]

On Thu, Jan 07, 2021 at 03:19:20PM +0800, Nelson Chu wrote:
> XPASS: ld-elf/compressed1d
> 
> I think the RISC-V big-endian support won't break the compress rules,
> just break the compressed1d test by accident.  Since the sizes of
> compressed and uncompressed .debug_aranges are the same for the
> riscv[32|64]be, but different for the little endian targets...  Jim
> already had written an alternative test compressed1d-alt.  Therefore,
> should we need to only xfail the riscv "little endian" targets for
> ld-elf/compressed1d, but allow the big endian passes?

Yes, that sounds like the right thing to do.

-- 
Alan Modra
Australia Development Lab, IBM

Re: [PATCH] ld: xfail riscv64be-*-* for ld-scripts/empty-address-2 tests.

[Nelson Chu]

On Thu, Jan 7, 2021 at 7:58 PM Alan Modra <amodra@gmail.com> wrote:
>
> On Thu, Jan 07, 2021 at 03:19:20PM +0800, Nelson Chu wrote:
> > XPASS: ld-elf/compressed1d
> >
> > I think the RISC-V big-endian support won't break the compress rules,
> > just break the compressed1d test by accident.  Since the sizes of
> > compressed and uncompressed .debug_aranges are the same for the
> > riscv[32|64]be, but different for the little endian targets...  Jim
> > already had written an alternative test compressed1d-alt.  Therefore,
> > should we need to only xfail the riscv "little endian" targets for
> > ld-elf/compressed1d, but allow the big endian passes?
>
> Yes, that sounds like the right thing to do.

Thanks!  I have sent the patch that tries to only xfail the riscv
little endian targets, but allow other targets to pass, including
riscv big endian targets.

Thanks
Nelson