Re: [PATCH] Add support for non-contiguous memory regions

[Christophe Lyon <christophe dot lyon at linaro dot org>]

On Wed, 19 Feb 2020 at 23:23, Alan Modra <amodra@gmail.com> wrote:
>
> On Wed, Feb 19, 2020 at 01:28:23PM +0100, Christophe Lyon wrote:
> > On Wed, 19 Feb 2020 at 08:19, Alan Modra <amodra@gmail.com> wrote:
> > >
> > > On Fri, Feb 14, 2020 at 02:53:55PM +0100, Christophe Lyon wrote:
> > > > Here are the updated patches.
> > >
> > > Wrong set of patches?  elfnn-aarch64.c, elf32-metag.c, and
> > > elf32-nios2.c won't compile with these patches applied.  All due to
> > > missing "struct bfd_link_info *info;" in build_one_stub.
> > >
> >
> > Oops, thanks for noticing.
> > I don't understand what happened. Here's the right version (the only
> > difference is the declaration of struct bfd_link_info *info; in
> > build_one_stub.
>

Hi Alan,

Thanks for the feedback.

> The test results don't inspire me with confidence.  A lot of the fails
> are due to targets not supporting -mlittle-endian, but there are other
> fails that show the patch series needs more work before it is ready
> for review.  Also, please don't write tests that require one
> particular endianness.  Write the objdump match patterns to suit both
> endians instead.  Something like this:
>
> Contents of section \.raml:
>  1fff0000 (010+ 020+ 030+|0+01 0+02 0+03)  .*
>
OK, I was wondering/hoping there's a nicer way of achieving this. I'll
update my patch to use this pattern.

>
> alpha-linux  FAIL: non-contiguous
> alpha-netbsd  FAIL: non-contiguous
> alpha-unknown-freebsd4.7  FAIL: non-contiguous
> am33_2.0-linux  FAIL: non-contiguous
> arc-elf  FAIL: non-contiguous
> arc-linux-uclibc  FAIL: non-contiguous
> arm-nacl  FAIL: non-contiguous-arm2
> arm-nacl  FAIL: non-contiguous-arm3
> arm-nacl  FAIL: non-contiguous-arm5
> arm-nacl  FAIL: non-contiguous-arm6
> bfin-elf  FAIL: non-contiguous
> bfin-linux-uclibc  FAIL: non-contiguous
> cr16-elf  FAIL: non-contiguous
> cris-elf  FAIL: non-contiguous
> cris-linux  FAIL: non-contiguous
> crisv32-linux  FAIL: non-contiguous
> crx-elf  FAIL: non-contiguous
> d10v-elf  FAIL: non-contiguous
> epiphany-elf  FAIL: non-contiguous
> fr30-elf  FAIL: non-contiguous
> frv-elf  FAIL: non-contiguous
> frv-linux  FAIL: non-contiguous
> ft32-elf  FAIL: non-contiguous
> h8300-elf  FAIL: non-contiguous
> h8300-linux  FAIL: non-contiguous
> i386-lynxos  FAIL: non-contiguous
> i586-linux  FAIL: non-contiguous
> i686-nacl  FAIL: non-contiguous
> i686-nto  FAIL: non-contiguous
> i686-pc-elf  FAIL: non-contiguous
> ia64-elf  FAIL: non-contiguous
> ia64-freebsd5  FAIL: non-contiguous
> ia64-linux  FAIL: non-contiguous
> ia64-netbsd  FAIL: non-contiguous
> ip2k-elf  FAIL: non-contiguous
> iq2000-elf  FAIL: non-contiguous
> lm32-elf  FAIL: non-contiguous
> lm32-linux  FAIL: non-contiguous
> m32c-elf  FAIL: non-contiguous
> m32r-elf  FAIL: non-contiguous
> m32r-linux  FAIL: non-contiguous
> m68k-elf  FAIL: non-contiguous
> m68k-linux  FAIL: non-contiguous
> mcore-elf  FAIL: non-contiguous
> mep-elf  FAIL: non-contiguous
> microblaze-elf  FAIL: non-contiguous
> microblaze-linux  FAIL: non-contiguous
> mips64el-openbsd  FAIL: non-contiguous
> mips64-linux  FAIL: non-contiguous
> mips64-openbsd  FAIL: non-contiguous
> mipsel-linux-gnu  FAIL: non-contiguous
> mipsisa32el-linux  FAIL: non-contiguous
> mips-linux  FAIL: non-contiguous
> mips-sgi-irix6  FAIL: non-contiguous
> mipstx39-elf  FAIL: non-contiguous
> mn10200-elf  FAIL: non-contiguous
> mn10300-elf  FAIL: non-contiguous
> moxie-elf  FAIL: non-contiguous
> msp430-elf  FAIL: non-contiguous
> mt-elf  FAIL: non-contiguous
> nds32be-elf  FAIL: non-contiguous
> nds32le-linux  FAIL: non-contiguous
> or1k-elf  FAIL: non-contiguous
> or1k-linux  FAIL: non-contiguous
> powerpc64-freebsd  FAIL: non-contiguous-powerpc
> powerpc64le-linux  FAIL: non-contiguous-powerpc
> powerpc64-linux  FAIL: non-contiguous-powerpc
> powerpc-freebsd  FAIL: non-contiguous
> pru-elf  FAIL: non-contiguous
> riscv32-elf  FAIL: non-contiguous
> riscv64-linux  FAIL: non-contiguous
> rl78-elf  FAIL: non-contiguous
> s390-linux  FAIL: non-contiguous
> s390x-linux  FAIL: non-contiguous
> score-elf  FAIL: non-contiguous
> shle-unknown-netbsdelf  FAIL: non-contiguous
> sh-linux  FAIL: non-contiguous
> sh-nto  FAIL: non-contiguous
> sh-rtems  FAIL: non-contiguous
> sparc64-linux  FAIL: non-contiguous
> sparc-elf  FAIL: non-contiguous
> sparc-linux  FAIL: non-contiguous
> sparc-sun-solaris2  FAIL: non-contiguous
> spu-elf  FAIL: non-contiguous
> tic6x-elf  FAIL: non-contiguous
> tilegx-linux  FAIL: non-contiguous
> tilepro-linux  FAIL: non-contiguous
> v850-elf  FAIL: non-contiguous
> vax-netbsdelf  FAIL: non-contiguous
> visium-elf  FAIL: non-contiguous
> x86_64-cloudabi  FAIL: non-contiguous
> x86_64-linux  FAIL: non-contiguous
> x86_64-nacl  FAIL: non-contiguous
> x86_64-pc-linux-gnux32  FAIL: non-contiguous
> xc16x-elf  FAIL: non-contiguous
> xstormy16-elf  FAIL: non-contiguous
> xtensa-elf  FAIL: non-contiguous
> z80-elf  FAIL: non-contiguous
>

So... I did run the tests on a few targets, but obviously not this long list.
How do you manage this? Is that list maintained somewhere?
Are there scripts to exercise all these targets? (--enable-targets=all
is not enough AFAICT)

Or should I just copy & paste that list from your email and
for i in <that-list>
do
  configure --target=$i
  make all check
done

Thanks,

Christophe


> --
> Alan Modra
> Australia Development Lab, IBM