error unpacking newlib-1.19.0: not in gzip format

[Poupapaa]

Hi!

I cloned the toolchain today and am getting an error while the script tries to unpack newlib-1.19.0, saying it's not in gzip format.

Steps taken:
git clone git://github.com/esden/summon-arm-toolchain
cd summon-arm-toolchain
git checkout remotes/origin/dev
./summon-arm-toolchain PREFIX=/usr/local/summon-arm-toolchain SUDO=sudo OOCD_EN=0 LIBSTM32_EN=1

(...)
gcc-linaro-4.6-2011.10/libiberty/lrealpath.c
gcc-linaro-4.6-2011.10/libiberty/memcpy.c

---

• Unpacking newlib-1.19.0

  ---

  archive type gz

gzip: stdin: not in gzip format
tar: Child returned status 1
tar: Error is not recoverable: exiting now

Running Linux Mint 14 Cinnamon, kernel 3.5.0-17-generic

[Piotr Esden-Tempski]

Thanks for sending this in. The dev branch is currently obsolete. Please try again with the master branch.

[Poupapaa]

Thank you for your response esden.
I tried the master branch issuing git checkout remotes/origin/master. Now make install fails for libgcc.

Trace:
(...)
chmod 644 /usr/local/summon-arm-toolchain/lib/gcc/arm-none-eabi/4.7.3/thumb/cortex-m4/float-abi-hard/fpuv4-sp-d16/libgcov.a
/usr/local/summon-arm-toolchain/arm-none-eabi/bin/ranlib /usr/local/summon-arm-toolchain/lib/gcc/arm-none-eabi/4.7.3/thumb/cortex-m4/float-abi-hard/fpuv4-sp-d16/libgcov.a
parts="crtbegin.o crtend.o crti.o crtn.o";
for file in $parts; do
rm -f /usr/local/summon-arm-toolchain/lib/gcc/arm-none-eabi/4.7.3/thumb/cortex-m4/float-abi-hard/fpuv4-sp-d16/$file;
/usr/bin/install -c -m 644 $file /usr/local/summon-arm-toolchain/lib/gcc/arm-none-eabi/4.7.3/thumb/cortex-m4/float-abi-hard/fpuv4-sp-d16/;
case $file in
.a)
/usr/local/summon-arm-toolchain/arm-none-eabi/bin/ranlib ../../../../../.././gcc/thumb/cortex-m4/float-abi-hard/fpuv4-sp-d16/$file ;;
esac;
done
make[4]: Leaving directory /home/sven/Documents/work/arm/cortex-m4/summon-arm-toolchain/build/arm-none-eabi/thumb/cortex-m4/float-abi-hard/fpuv4-sp-d16/libgcc' make[3]: Leaving directory/home/sven/Documents/work/arm/cortex-m4/summon-arm-toolchain/build/arm-none-eabi/libgcc'
make[2]: Leaving directory /home/sven/Documents/work/arm/cortex-m4/summon-arm-toolchain/build/arm-none-eabi/libgcc' make[1]: Leaving directory/home/sven/Documents/work/arm/cortex-m4/summon-arm-toolchain/build'
make: ** [install] Error 2

[Piotr Esden-Tempski]

That is a strange error because I don't see the actual line that states what the error was. I think I had something similar and it was gone when I rerun the compile. You could also go into the build directory after the error and run make in there. Then you should be able to see the source of the error itself.

[Poupapaa]

Yes, it is strange indeed.
I just re-ran the script, and now it has passed where it failed before, but I get a new error on libcmsis.

Trace:
(...)
libcmsis-379ad8eb38feb6703f2f5f2e457742b7e09e3b2c/include/cmsis/system_stm32.h
libcmsis-379ad8eb38feb6703f2f5f2e457742b7e09e3b2c/include/cmsis/core_cm3.h
libcmsis-379ad8eb38feb6703f2f5f2e457742b7e09e3b2c/include/cmsis/stm32.h

---

• Building libcmsis-379ad8eb38feb6703f2f5f2e457742b7e09e3b2c

  ---

  make -C src all
  make[1]: Entering directory /home/sven/Documents/work/arm/cortex-m4/summon-arm-toolchain/libcmsis-379ad8eb38feb6703f2f5f2e457742b7e09e3b2c/src' arm-none-eabi-gcc -I../include -Wall -ansi -std=c99 -fno-common -Os -g -mcpu=cortex-m3 -mthumb -ffunction-sections -fdata-sections -c -o core_cm3.o core_cm3.c /tmp/cc8O4ejE.s: Assembler messages: /tmp/cc8O4ejE.s:452: Error: registers may not be the same --strexb r0,r0,[r1]'
  /tmp/cc8O4ejE.s:477: Error: registers may not be the same -- strexh r0,r0,[r1]' make[1]: *** [core_cm3.o] Error 1 make[1]: Leaving directory/home/sven/Documents/work/arm/cortex-m4/summon-arm-toolchain/libcmsis-379ad8eb38feb6703f2f5f2e457742b7e09e3b2c/src'
  make: *** [all] Error 2

Any ideas? Re-running for a third time doesn't seem to fix this one.

[Poupapaa]

Seems to be a known bug that has been patched 11 months ago. Please have a look at stlink-org/stlink#65. Maybe you can pull in the patch?

[Piotr Esden-Tempski]

The previous bug with having to rerun the script is something I can observe here too. Where linaro gcc fails to compile on the first try. Not sure how to solve it yet.

I added the patch you pointed out to libcmsis. Still I would encourage you to use libopencm3 instead of the st libraries in the future. :)