RISC-V GNU Compiler Toolchain

1,176 views
Skip to first unread message

J Osmany

unread,
Oct 25, 2016, 3:55:35 AM10/25/16
to sw-...@groups.riscv.org, sw-...@lists.riscv.org

Hello


From https://github.com/riscv/riscv-gnu-toolchain, i have followed the instructions for "Installation (Newlib)"

and have managed to install the RISC-V C and C++ compiler toolchain.


Next i wanted to run the DejaGnu test suite. Again followed the instructions:


./configure -prefix=$RISCV -disable-float -disable-linux

make check-gcc-newlib


However, the make is failing because it cannot find the "runtest" command.


Log from running the above make command:


bash-4.1$ pwd
/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain
bash-4.1$  make check-gcc-newlib
export PATH=/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/opt_riscv/bin:/awk:/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/opt_riscv/bin:/vl/edatools/eda/bin/wrappers:/vl/edatools/eda/bin:/vl/edatools/bin:/usr/local/bin:/bin:/usr/bin:.:vl/edatools/modules/tools:/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/opt_riscv/bin && \
        export DEJAGNULIBS=/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/riscv-dejagnu && \
        cd build-gcc-newlib && \
        /vl/edatools/tools/make/Linux/bin/make check-gcc "RUNTESTFLAGS=--target_board=riscv-sim"
make[1]: Entering directory `/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/build-gcc-newlib'
make[2]: Entering directory `/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/build-gcc-newlib/gcc'
rm -rf testsuite/gcc-parallel
make[3]: Entering directory `/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/build-gcc-newlib/gcc'
(rootme=`${PWDCMD-pwd}`; export rootme; \
        srcdir=`cd /user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/src/newlib-gcc/gcc; ${PWDCMD-pwd}` ; export srcdir ; \
        if [ -n "" ] \
           && [ -n "$GCC_RUNTEST_PARALLELIZE_DIR" ] \
           && [ -f testsuite/gcc-parallel/finished ]; then \
          rm -rf testsuite/gcc; \
        else \
          cd testsuite/gcc; \
          rm -f tmp-site.exp; \
          sed '/set tmpdir/ s|testsuite$|testsuite/gcc|' \
                < ../../site.exp > tmp-site.exp; \
          /bin/sh ${srcdir}/../move-if-change tmp-site.exp site.exp; \
          EXPECT=`if [ -f ${rootme}/../expect/expect ] ; then echo ${rootme}/../expect/expect ; else echo expect ; fi` ; export EXPECT ; \
          if [ -f ${rootme}/../expect/expect ] ; then  \
            TCL_LIBRARY=`cd .. ; cd ${srcdir}/../tcl/library ; ${PWDCMD-pwd}` ; \
            export TCL_LIBRARY ; \
          fi ; \
          `if [ -f ${srcdir}/../dejagnu/runtest ] ; then echo ${srcdir}/../dejagnu/runtest ; else echo runtest; fi` --tool gcc --target_board=riscv-sim; \
          if [ -n "$GCC_RUNTEST_PARALLELIZE_DIR" ] ; then \
            touch ${rootme}/testsuite/gcc-parallel/finished; \
          fi ; \
        fi )
/bin/sh: line 17: runtest: command not found
make[3]: Leaving directory `/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/build-gcc-newlib/gcc'
rm -rf testsuite/g++-parallel
make[3]: Entering directory `/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/build-gcc-newlib/gcc'
(rootme=`${PWDCMD-pwd}`; export rootme; \
        srcdir=`cd /user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/src/newlib-gcc/gcc; ${PWDCMD-pwd}` ; export srcdir ; \
        if [ -n "" ] \
           && [ -n "$GCC_RUNTEST_PARALLELIZE_DIR" ] \
           && [ -f testsuite/g++-parallel/finished ]; then \
          rm -rf testsuite/g++; \
        else \
          cd testsuite/g++; \
         
rm -f tmp-site.exp; \
          sed '/set tmpdir/ s|testsuite$|testsuite/g++|' \
                < ../../site.exp > tmp-site.exp; \
          /bin/sh ${srcdir}/../move-if-change tmp-site.exp site.exp; \
          EXPECT=`if [ -f ${rootme}/../expect/expect ] ; then echo ${rootme}/../expect/expect ; else echo expect ; fi` ; export EXPECT ; \
          if [ -f ${rootme}/../expect/expect ] ; then  \
            TCL_LIBRARY=`cd .. ; cd ${srcdir}/../tcl/library ; ${PWDCMD-pwd}` ; \
            export TCL_LIBRARY ; \
          fi ; \
          `if [ -f ${srcdir}/../dejagnu/runtest ] ; then echo ${srcdir}/../dejagnu/runtest ; else echo runtest; fi` --tool g++ --target_board=riscv-sim; \
          if [ -n "$GCC_RUNTEST_PARALLELIZE_DIR" ] ; then \
            touch ${rootme}/testsuite/g++-parallel/finished; \
          fi ; \
        fi )
/bin/sh: line 17: runtest: command not found

Looking at the Makefile in build-gcc-newlib/gcc, at line 796 we have


RUNTEST = `if [ -f $${srcdir}/../dejagnu/runtest ] ; then \
           echo $${srcdir}/../dejagnu/runtest ; \
        else echo runtest; fi`


But ${srcdir}/../dejagnu/ does not exist.


So i hacked the above to


RUNTEST = `if [ -f $${srcdir}/../../../riscv-dejagnu/runtest ] ; then \
           echo $${srcdir}/../../../riscv-dejagnu/runtest ; \
        else echo runtest; fi`


However, re-running the make command again did not work. Getting the following from running the make command:

bash-4.1$ pwd
/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain
bash-4.1$  make check-gcc-newlib
export PATH=/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/opt_riscv/bin:/awk:/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/opt_riscv/bin:/vl/edatools/eda/bin/wrappers:/vl/edatools/eda/bin:/vl/edatools/bin:/usr/local/bin:/bin:/usr/bin:.:vl/edatools/modules/tools:/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/opt_riscv/bin && \
        export DEJAGNULIBS=/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/riscv-dejagnu && \
        cd build-gcc-newlib && \
        /vl/edatools/tools/make/Linux/bin/make check-gcc "RUNTESTFLAGS=--target_board=riscv-sim"
make[1]: Entering directory `/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/build-gcc-newlib'
make[2]: Entering directory `/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/build-gcc-newlib/gcc'
Making a new config file...
echo "set tmpdir /user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/build-gcc-newlib/gcc/testsuite" >> ./site.tmp
rm -rf testsuite/gcc-parallel
make[3]: Entering directory `/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/build-gcc-newlib/gcc'
(rootme=`${PWDCMD-pwd}`; export rootme; \
        srcdir=`cd /user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/src/newlib-gcc/gcc; ${PWDCMD-pwd}` ; export srcdir ; \
        if [ -n "" ] \
           && [ -n "$GCC_RUNTEST_PARALLELIZE_DIR" ] \
           && [ -f testsuite/gcc-parallel/finished ]; then \
          rm -rf testsuite/gcc; \
        else \
          cd testsuite/gcc; \
          rm -f tmp-site.exp; \
          sed '/set tmpdir/ s|testsuite$|testsuite/gcc|' \
                < ../../site.exp > tmp-site.exp; \
          /bin/sh ${srcdir}/../move-if-change tmp-site.exp site.exp; \
          EXPECT=`if [ -f ${rootme}/../expect/expect ] ; then echo ${rootme}/../expect/expect ; else echo expect ; fi` ; export EXPECT ; \
          if [ -f ${rootme}/../expect/expect ] ; then  \
            TCL_LIBRARY=`cd .. ; cd ${srcdir}/../tcl/library ; ${PWDCMD-pwd}` ; \
            export TCL_LIBRARY ; \
          fi ; \
          `if [ -f ${srcdir}/../../../riscv-dejagnu/runtest ] ; then echo ${srcdir}/../../../riscv-dejagnu/runtest ; else echo runtest; fi` --tool gcc --target_board=riscv-sim; \
          if [ -n "$GCC_RUNTEST_PARALLELIZE_DIR" ] ; then \
            touch ${rootme}/testsuite/gcc-parallel/finished; \
          fi ; \
        fi )
ERROR: unable to find expect on the PATH
make[3]: Leaving directory `/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/build-gcc-newlib/gcc'
rm -rf testsuite/g++-parallel
make[3]: Entering directory `/user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/build-gcc-newlib/gcc'
(rootme=`${PWDCMD-pwd}`; export rootme; \
        srcdir=`cd /user/jdo.tmp2/risc-v/sw_tools_newlib_18_Oct_2016/riscv-gnu-toolchain/src/newlib-gcc/gcc; ${PWDCMD-pwd}` ; export srcdir ; \
        if [ -n "" ] \
           && [ -n "$GCC_RUNTEST_PARALLELIZE_DIR" ] \
           && [ -f testsuite/g++-parallel/finished ]; then \
          rm -rf testsuite/g++; \
        else \
          cd testsuite/g++; \
          rm -f tmp-site.exp; \
          sed '/set tmpdir/ s|testsuite$|testsuite/g++|' \
                < ../../site.exp > tmp-site.exp; \
          /bin/sh ${srcdir}/../move-if-change tmp-site.exp site.exp; \
          EXPECT=`if [ -f ${rootme}/../expect/expect ] ; then echo ${rootme}/../expect/expect ; else echo expect ; fi` ; export EXPECT ; \
          if [ -f ${rootme}/../expect/expect ] ; then  \
            TCL_LIBRARY=`cd .. ; cd ${srcdir}/../tcl/library ; ${PWDCMD-pwd}` ; \
            export TCL_LIBRARY ; \
          fi ; \
          `if [ -f ${srcdir}/../../../riscv-dejagnu/runtest ] ; then echo ${srcdir}/../../../riscv-dejagnu/runtest ; else echo runtest; fi` --tool g++ --target_board=riscv-sim; \
          if [ -n "$GCC_RUNTEST_PARALLELIZE_DIR" ] ; then \
            touch ${rootme}/testsuite/g++-parallel/finished; \
          fi ; \
        fi )
ERROR: unable to find expect on the PATH


Therefore, suspect my hack to RUNTEST was not correct.

Any pointers to resolve the above issue much appreciated.




README.md RISC-V GNU Compiler Toolchain. This is the RISC-V C and C++ cross-compiler. It supports two build modes: a generic ELF/Newlib toolchain and a more ...


Best Regards

J.Osmany

Kito Cheng

unread,
Oct 25, 2016, 5:25:54 AM10/25/16
to J Osmany, sw-...@groups.riscv.org, sw-...@lists.riscv.org
runtest is provide by DejaGnu Testing Framework,
you can install it by package system for example:

$ sudo apt-get install dejagnu # for Ubuntu/Debian
$ sudo dnf install dejagnu # For Fedora
$ sudo yum install dejagnu # For CentOS
...


--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+unsubscribe@groups.riscv.org.
To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/VI1PR0501MB236516F5CC35FFEC92951FC8EBA80%40VI1PR0501MB2365.eurprd05.prod.outlook.com.

J Osmany

unread,
Oct 25, 2016, 7:27:22 AM10/25/16
to Kito Cheng, sw-...@groups.riscv.org

Thanks for the pointer.



Best Regards

J.Osmany



From: npic...@gmail.com <npic...@gmail.com> on behalf of Kito Cheng <kito....@gmail.com>
Sent: 25 October 2016 10:25
To: J Osmany
Cc: sw-...@groups.riscv.org; sw-...@lists.riscv.org
Subject: Re: [sw-dev] RISC-V GNU Compiler Toolchain
 
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+un...@groups.riscv.org.

To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.

Sober Liu

unread,
Dec 5, 2016, 11:05:12 AM12/5/16
to J Osmany, Kito Cheng, sw-...@groups.riscv.org
Hi,
Does anyone run dejagnu for riscv-gcc recently with master branch?
I synced a new version today from master branch.
When I build with default ./build.sh, the whole riscv-tools built success. Then I run “make-gcc-newlib” and see many FAILs from unknown insn error. I think it’s known issue as gdb simulator do not support FP insn.
When I build with soft-float configure as mentioned in github page, “make newlib” failed when build riscv-pk. It tells “…/6.1.0/crtbegin.o: can't link hard-float modules with soft-float modules”. I think this is also a known issue but should has been fixed around in May? I see another mail thread talking about the gcc option changes around msoft-float. But I am not sure whether related changes also should be applied on gcc configure to get library in soft float mode.

So, what's the expected steps to configure/build, run dejaGnu and what's the expected results?
If it's still expected to pass, I can go on to check setting from my own side.
Thanks a lot.

Copy words from https://github.com/riscv/riscv-gnu-toolchain/tree/55f83087a711a24b8704309b0dd32e6071fcafad here:
./configure --prefix=$RISCV --disable-float --disable-linux
make newlib
make check-gcc-newlib

From: J Osmany [mailto:liv...@hotmail.co.uk]
Sent: Tuesday, October 25, 2016 7:27 PM
To: Kito Cheng
Cc: sw-...@groups.riscv.org
Subject: Re: [sw-dev] RISC-V GNU Compiler Toolchain

Thanks for the pointer.


Best Regards

J.Osmany

________________________________________
GitHub - riscv/riscv-gnu-toolchain: GNU toolchain for RISC ...
github.com
README.md RISC-V GNU Compiler Toolchain. This is the RISC-V C and C++ cross-compiler. It supports two build modes: a generic ELF/Newlib toolchain and a more ...


Best Regards

J.Osmany
--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+un...@groups.riscv.org.
To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/VI1PR0501MB236516F5CC35FFEC92951FC8EBA80%40VI1PR0501MB2365.eurprd05.prod.outlook.com.

--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+un...@groups.riscv.org.
To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/CA%2ByXCZBnofKijacodHSnGdKWwhoVgvVNdEKFZ%3D7Nrwui_Z52Qw%40mail.gmail.com.
--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+un...@groups.riscv.org.
To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/VI1PR0501MB23656D1866D9C471ABF4358DEBA80%40VI1PR0501MB2365.eurprd05.prod.outlook.com.

-----------------------------------------------------------------------------------
This email message is for the sole use of the intended recipient(s) and may contain
confidential information. Any unauthorized review, use, disclosure or distribution
is prohibited. If you are not the intended recipient, please contact the sender by
reply email and destroy all copies of the original message.
-----------------------------------------------------------------------------------

Sober Liu

unread,
Dec 7, 2016, 11:15:29 AM12/7/16
to J Osmany, Kito Cheng, sw-...@groups.riscv.org
I find that crt0.o is the file with hard-float ABI when gcc is configured with soft-float (./configure --prefix=$RISCV --disable-float --disable-linux).
This file comes from libgloss. Generated from riscv-gnu-toolchain/build/build-gcc-newlib/riscv64-unknown-elf/libgloss/riscv

In this path we can see another file syscall.o is in soft-float ABI.
> riscv-gnu-toolchain/build/riscv-build/bin/riscv64-unknown-elf-readelf -a crt0.o | grep Flags:
Flags: 0x0
> riscv-gnu-toolchain/build/riscv-build/bin/riscv64-unknown-elf-readelf -a syscalls.o | grep Flags:
Flags: 0x2, soft-float ABI

From log file we see they are using same CFLAGS and xgcc cmd for compile.
Adding gcc option -save-temps -v, we can see different comes from .s file.
There is a line in syscall.s: ".option soft-float" and no find in crt0.s.
Additional, this line comes from riscv.c: riscv_file_start().
My understanding is this function called only when cc1 invoked by gcc to compile .c->.s .

For .S file gcc invoke its internal preprocessor and failed to generate such string in .s.
E.g., riscv-gnu-toolchain/build/build-gcc-newlib/./gcc/as -v -I . -I /home/scratch.soberl_maxwell/arch1/20161205/riscv-tools/riscv-gnu-toolchain/build/src/newlib-gcc/libgloss/riscv --traditional-format -m64 -o crt0.o crt0.s
So I think to compile .S file with soft-float, either let processor to append similar line to .s, or let gcc to forward a parameter to 'as'.

BTW, I am work with master branch clone on 20161205.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/bc8c0849d1e24cee9623266a51dffe5f%40HKMAIL101.nvidia.com.

Tommy Murphy

unread,
Dec 7, 2016, 11:53:38 AM12/7/16
to RISC-V SW Dev, liv...@hotmail.co.uk, kito....@gmail.com
I don't know if it's relevant but I'm also having problems trying to get the tools build and working with soft float support.
Here's my original issue report with an update from earlier today appended:


BTW - when I run the test suite (make check-gcc-newlib) I also get many FAIL messages although I can't see any details about why the relevant tests fail.
Should the test suite run with no FAIL messages at all or is there some acceptable level of failures?

Sober Liu

unread,
Dec 7, 2016, 12:01:18 PM12/7/16
to Tommy Murphy, RISC-V SW Dev, liv...@hotmail.co.uk, kito....@gmail.com

Inline.

 

From: Tommy Murphy [mailto:tommy_...@hotmail.com]
Sent: Thursday, December 08, 2016 12:54 AM
To: RISC-V SW Dev
Cc: liv...@hotmail.co.uk; kito....@gmail.com
Subject: Re: [sw-dev] RISC-V GNU Compiler Toolchain

 

I don't know if it's relevant but I'm also having problems trying to get the tools build and working with soft float support.

Here's my original issue report with an update from earlier today appended:

 

https://github.com/riscv/riscv-gnu-toolchain/issues/199

[sober] It’s not  same issue. Here are something bad in compiler (cc1). And my issue comes from linker (ld) that tells “can't link hard-float modules with soft-float modules”

 

BTW - when I run the test suite (make check-gcc-newlib) I also get many FAIL messages although I can't see any details about why the relevant tests fail.

Should the test suite run with no FAIL messages at all or is there some acceptable level of failures?

[sober] You could find details in riscv-gnu-toolchain/build/build-gcc-newlib/gcc/testsuite/gcc/gcc.log.
And you may also modify this line by add several “-v” in riscv-gnu-toolchain/Makefile.in: $(MAKE) check-gcc "RUNTESTFLAGS=--target_board=riscv-sim -v -v"

 

--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+un...@groups.riscv.org.
To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.

This email message is for the sole use of the intended recipient(s) and may contain confidential information.  Any unauthorized review, use, disclosure or distribution is prohibited.  If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message.

Sober Liu

unread,
Dec 8, 2016, 10:46:42 AM12/8/16
to J Osmany, Kito Cheng, sw-...@groups.riscv.org
After make a WAR to force ‘as’ to generate soft ABI always to run gcc regression, I see many case failed with “program stopped with signal 4 (Illegal instruction).”
The directed reason is fsd insn generated; and gdb simulator believe it belongs to unsupported D extension.

Take strcpy-chk.x8 execution as an example, in strcpy-chk.x8.ltrans0.s I see following lines:
...
.option soft-float
...
fsd fs0,168(sp)
fsd fs1,160(sp)
...
This is a little bit confusing. Although -msoft-float contained in CFLAGS, default ARCH is RV64IMAFD. So gcc believe these reg/insn could be used.

Finally I add -march-RV64IMA to cflags for test and this case works.
Then we have alternative choices: To add this option to cflags defined in riscvi-sim.exp or to configure gcc with --with-arch=RV64IMA.
I prefer to change riscv-sim.exp.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/a3af7374e0bd4e4d92d3d52538bfb32c%40HKMAIL101.nvidia.com.

Kito Cheng

unread,
Dec 8, 2016, 9:17:28 PM12/8/16
to Sober Liu, J Osmany, sw-...@groups.riscv.org
You should configure gcc with --with-arch=RV64IMA instead of edit riscvi-sim.exp since newlib and libgcc also need compile with right arch, otherwise the library code may contain F or D instructions.
 

To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+unsubscribe@groups.riscv.org.

To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/VI1PR0501MB236516F5CC35FFEC92951FC8EBA80%40VI1PR0501MB2365.eurprd05.prod.outlook.com.

--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+unsubscribe@groups.riscv.org.

To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/CA%2ByXCZBnofKijacodHSnGdKWwhoVgvVNdEKFZ%3D7Nrwui_Z52Qw%40mail.gmail.com.
--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+unsubscribe@groups.riscv.org.

To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/VI1PR0501MB23656D1866D9C471ABF4358DEBA80%40VI1PR0501MB2365.eurprd05.prod.outlook.com.

-----------------------------------------------------------------------------------
This email message is for the sole use of the intended recipient(s) and may contain confidential information.  Any unauthorized review, use, disclosure or distribution is prohibited.  If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message.
-----------------------------------------------------------------------------------

--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+unsubscribe@groups.riscv.org.

To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/bc8c0849d1e24cee9623266a51dffe5f%40HKMAIL101.nvidia.com.

--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+unsubscribe@groups.riscv.org.
--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+unsubscribe@groups.riscv.org.

To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.

Sober Liu

unread,
Dec 8, 2016, 11:05:06 PM12/8/16
to Kito Cheng, J Osmany, sw-...@groups.riscv.org
Hi Kito,
Yes, you are right that we should use configure option --with-arch=RV64IMA to make sure library side.
And it will be nice if gcc could warning on conflict between IMAFD extension and -msoft-float option, or simply disable FD extension.
But anyway the more critical bug is to fix the msoft-float supports for .S file.

Now I got following result:
=== gcc Summary ===
# of expected passes 79483
# of unexpected failures 206
# of expected failures 100
# of unresolved testcases 71
# of unsupported tests 1659

I see most FAILs cause by following reason:
calling DUPFFexgcd on degrees 2 and 2
DUPFFexgcd called on degrees 2 and 2

Could it caused by privilege spec changes between 1.7 and 1.9?
I will look into details later. And I will try to use spike as the simulator locally.

Thanks.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+un...@groups.riscv.org.
To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/VI1PR0501MB236516F5CC35FFEC92951FC8EBA80%40VI1PR0501MB2365.eurprd05.prod.outlook.com.

--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+un...@groups.riscv.org.
To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/CA%2ByXCZBnofKijacodHSnGdKWwhoVgvVNdEKFZ%3D7Nrwui_Z52Qw%40mail.gmail.com.
--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+un...@groups.riscv.org.
To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/VI1PR0501MB23656D1866D9C471ABF4358DEBA80%40VI1PR0501MB2365.eurprd05.prod.outlook.com.

-----------------------------------------------------------------------------------
This email message is for the sole use of the intended recipient(s) and may contain confidential information.  Any unauthorized review, use, disclosure or distribution is prohibited.  If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message.
-----------------------------------------------------------------------------------

--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+un...@groups.riscv.org.
To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/bc8c0849d1e24cee9623266a51dffe5f%40HKMAIL101.nvidia.com.

--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+un...@groups.riscv.org.
To post to this group, send email to sw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/sw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/sw-dev/a3af7374e0bd4e4d92d3d52538bfb32c%40HKMAIL101.nvidia.com.

--
You received this message because you are subscribed to the Google Groups "RISC-V SW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sw-dev+un...@groups.riscv.org.
Reply all
Reply to author
Forward
0 new messages