| Message ID | 1559735815-20093-12-git-send-email-vincentc@andestech.com |
|---|---|
| State | New |
| Headers | show |
| Series | nds32 glibc port, v4 | expand |
On Wed, 5 Jun 2019, Vincent Chen wrote: > From: CheWei Chunang <cnoize@andestech.com> > > The nds32 toochain cannot be generated successfully by this patch because > the nds32 port of Binutils and GCC on upstream is too old. This patch is > used to help reviewer to review the glibc port for Andes nds32 ISA. For a new port to go into glibc, the upstream mainline versions of binutils / GCC / Linux kernel need to have all the features required to build the port and run tests with good results (even if release branches are too old to work well - though if it's just a matter of bug fixes, backporting those to release branches may be a good idea). Ports depending on non-upstream changes to those components can't be accepted (we had trouble in the past with MicroBlaze and NaCl depending on things that weren't upstream, fixed by getting some MicroBlaze GCC patches upstream and removing the NaCl port; we don't want to go there again).
On Fri, Jun 07, 2019 at 06:47:49AM +0800, Joseph Myers wrote: > On Wed, 5 Jun 2019, Vincent Chen wrote: > > > From: CheWei Chunang <cnoize@andestech.com> > > > > The nds32 toochain cannot be generated successfully by this patch because > > the nds32 port of Binutils and GCC on upstream is too old. This patch is > > used to help reviewer to review the glibc port for Andes nds32 ISA. > > For a new port to go into glibc, the upstream mainline versions of > binutils / GCC / Linux kernel need to have all the features required to > build the port and run tests with good results (even if release branches > are too old to work well - though if it's just a matter of bug fixes, > backporting those to release branches may be a good idea). Ports > depending on non-upstream changes to those components can't be accepted > (we had trouble in the past with MicroBlaze and NaCl depending on things > that weren't upstream, fixed by getting some MicroBlaze GCC patches > upstream and removing the NaCl port; we don't want to go there again). > Sorry, I did not find the commit log is out-of-day. Now, build-many-glibcs.py can use upstream Binutils-2.32, GCC-9.0.1 and this patchset to generate nds32 toolchain and pass the test. I will correct this commit log and append the test log in the next submission. Thanks, Vincent
diff --git a/scripts/build-many-glibcs.py b/scripts/build-many-glibcs.py index c5821df..4bde3fa 100755 --- a/scripts/build-many-glibcs.py +++ b/scripts/build-many-glibcs.py @@ -299,6 +299,18 @@ class Context(object): 'ccopts': '-mabi=32'}, {'variant': 'n64-nan2008-soft', 'ccopts': '-mabi=64'}]) + self.add_config(arch='nds32le', + os_name='linux-gnu', + variant='v3', + gcc_cfg=['--with-arch=v3', '--with-cpu=n13', + '--enable-default-relax=no', + '--disable-multilib']) + self.add_config(arch='nds32le', + os_name='linux-gnu', + variant='v3f', + gcc_cfg=['--with-arch=v3f', '--with-cpu=n13', + '--enable-default-relax=no', + '--disable-multilib']) self.add_config(arch='nios2', os_name='linux-gnu') self.add_config(arch='powerpc', @@ -1269,6 +1281,7 @@ class Config(object): 'm68k': 'm68k', 'microblaze': 'microblaze', 'mips': 'mips', + 'nds32': 'nds32', 'nios2': 'nios2', 'powerpc': 'powerpc', 's390': 's390',
From: CheWei Chunang <cnoize@andestech.com> The nds32 toochain cannot be generated successfully by this patch because the nds32 port of Binutils and GCC on upstream is too old. This patch is used to help reviewer to review the glibc port for Andes nds32 ISA. 2019-06-03 Vincent Chen <vincentc@andestech.com> 2019-06-03 CheWei Chuang <cnoize@andestech.com> * scripts/build-many-glibcs.py (Context): Add nds32 targets. --- scripts/build-many-glibcs.py | 13 +++++++++++++ 1 file changed, 13 insertions(+)