Summary:
* Enhance crosstool-ng build process.
Details:
1. Linaro crosstool-ng patches:
* Update and commit the patch to strip debug sections.
* rm *-cc.exe symblink for win32.
* enable gdb for win32 (lp:918479)
* Update README to add flip dependence for build.
2. Verify the following bugs and change status to "Fix Committed"
* lp:906729 baremetal libc headers and libraries are in inconsistent places
* lP:906077 binaries: baremetal include files are in the wrong place
* lP:894527 binaries: remove all symlinks from the Windows build
* lp:889984 binaries: should step across helper functions
* lp:906662 binaries: pgversion can get out of sync with product
3. Analyzed bugs
* lp:915137 Link failure due to absolute paths: Need more
information to reproduce.
* lp:p18478 wind32 doesn't have pkg-config: Try to build pkg-config
for mingw32 host. But failed since it required preinstalled glib2. And
configure error if cross building the glib.
* lp:894528 binaries: can't run if the patch contains whitespace.
Can not reproduce it on cygwin, but still have issue in DOS cmdline
4. Verify the prerelease binary package on windows. And find and
workaround a gdb document name issue since filename on windows is case
insensitive (lp:909195).
5. Raise the gcc trunk build fail issue to crosstool-ng mail-list. But
not confirm "-EL" is the final root cause. Need more investigation.
Planned Absence:
* Jan 21-29, 2012 Chinese Spring Festival
Best regards!
-Zhenqiang
Hello experts,
I don't know whether this is the place to ask this question.Please forgive
if i am wrong.
I am trying to cross-compile Eglibc,(which i thought is similar to the
glibc) library with Linaro tool chain.My aim is to cross-compile the
library and add it with the tool-chain( i assumed that Linaro doesn't ahve
the eglibc libraries by itself).
.
I referred to the link http://www.eglibc.org/archives/patches/msg00078.html
(README.cross-compiling of the libc) and did the following.
I thought that binutils is not required as the tool chain itself contains
the loader, assembler etc.(which are the most important in the binutils).So
i skipped that step.(i am not sure whether this is right.please correct me
if i am wrong).
I did the below steps as in the above link.
1) Copied the Linux Kernel Headers
2) EGLIBC Headers and Preliminary Objects
2.1) configuring the Eglibc headers by the below steps.
appan@oplt$ ~/cross_compile_eglibc/ppc/obj/eglibc-headers$ BUILD_CC=gcc
CC=/home/appan/cross_compile_eglibc/ppc/tools/bin/arm-linux-gnueabi-gcc
CXX=/home/appan/cross_compile_eglibc/ppc/tools/bin/arm-linux-gnueabi-cpp
AR=/home/appan/cross_compile_eglibc/ppc/tools/bin/arm-linux-gnueabi-ar
RANLIB=/home/appaan/cross_compile_eglibc/ppc/tools/bin/arm-linux-gnueabi-ranlib
/home/appan/cross_compile_eglibc/src/libc/configure
--prefix=/home/appan/CC_EGLIBC
--with-headers=/home/appan/cross_compile_eglibc/ppc/sysroot/usr/include/
--build=i686-pc-linux-gnu --host=arm-unknown-linux-gnu --enable-add-ons
--with-tls --with-__thread
2.2) compilation
make install-headers install_root=$sysroot install-bootstrap-headers=yes
But following is the error i get.
> /home/appan/cross_compile_eglibc/ppc/obj/eglibc-headers/Versions.v.iT<http://versions.v.it/>
In file included from <stdin>:1:0:
ports/sysdeps/arm/nptl/tls.h:48:3: error: #*error "TLS support is required.*
"
make[1]: *** No rule to make target `headers_install'. Stop.
make[1]: Leaving directory `/home/appan/cross_compile_eglibc/src/libc'
make: *** [headers_install] Error 2
I tried googling, but i did not get the solution for this..
please provide me with some suggestions and that will be of great help to
me...
thanks and regards
appan
Hi there!
I need some help with how to handle this bug:
https://bugs.launchpad.net/gcc-linaro/+bug/873977
It looks like an improvement suggestion more than a bug.
Regards
Åsa
*Automatically save preprocessed source on ICE*In order to help with bug
reporting, gcc could save the preprocessed source of the unit that caused
the ICE, so one can just pick it up and attach it to bug reports instead of
trying to reproduce it manually.
Hi Peter. I've had a poke about in qemu-linaro, OpenStack, and
libvirt to get a feel for the KVM integration work. My scratch notes
are here:
https://wiki.linaro.org/MichaelHope/Sandbox/QEMUA15GuestNotes
I'm feeling happy about the integration. The parts that I've seen
have good separation, are architecture neutral, and have some ARM
support. We can prototype using virtual x86 based masters and QEMU
TCG based compute instances.
I've seen a few bugs and assumptions: OpenStack defaults to virtio,
boot=hda, and fixed kernel args; and libvirt can't parse the the
qemu-linaro version string. There will be more.
I'll feed this into the KVM 1.0 task list and share it.
-- Michael
Hi there. Starting next week we'll switch to the Zip conference line. See:
https://wiki.linaro.org/Resources/ZipConferenceLine
under 1-719-457-1414 for a list of numbers. A selected few are:
UK 0 808 101 1146
Germany 0 800 181 9019
Sweden 02 079 7556
Brazil, Sao Paulo Local +55 11 5582 6534
Australia 1 800 105 680
China +400 148 5400
NZ 0800 451 015
The code is 763 804 0680. I've updated the meeting calendar and will
update the wiki page once Monday comes around. The moderator code is
on the toolchain internal page.
Amber, could you update the public event please?
Ramana, could you update the performance call for two weeks time?
Maybe move it to linaro-events so I can edit it.
-- Michael
Hi Michael,
So, I gave my best shot at:
https://bugs.launchpad.net/gcc-linaro/+bug/914703
Kind of similar to the one in your walk-through, but I couldn't actually
reproduce on the version reported.
Could you please check if this looks OK?
Regards
Åsa
Hi Åsa. I've triaged a bug and written down my notes on the way. See:
https://wiki.linaro.org/MichaelHope/Sandbox/TriageRunThrough
This one turned out to be easy but at least you have the basics.
Could you triage the bug for real and see what I've missed?
I'm unsatisfied with the result on that bug. The fault is in the
Ubuntu Natty compiler which we've long since fixed in Linaro GCC. It
would be nice if we could point the bug reporter to an updated GCC.
Ask Marcin and Matthias what they think.
-- Michael
Summary:
* Enhance crosstool-ng build process.
Details:
1. crosstool-ng patches:
* Reproduce the build process by following the README and enhance
the document.
* Update .config to strip all toolchain executables
* Replace harded-coded VERSION by reading config file.
* Remove host libiberty.a
* Patch on how to strip debug sections is in discussion.
2. Update the crosstool-ng based embedded toolchain build scripts
according to the review comments. And replace most of hard-coded
config by reading them from .config.
3. Unexpected build fail for gcc 4.7/trunk from crosstool-ng due to
unsupported ENDIAN_LDFLAG "-EL", which is used in "LDFLAGS_FOR_TARGET"
to build libstdc++. (LDFLAGS_FOR_TARGET seams not used in gcc 4.6).
Need more investigation.
Plan:
* Fix the binary build related bugs.
* Ramp-up on gcc.
Planned Absence:
* Jan 21-29, 2012 Chinese Spring Festival
Best regards!
-Zhenqiang
Hi,
I use pre-built version of linaro toolchain (got from http://people.linaro.org/~michaelh/incoming/binaries/) to build a uImage.
The build succeeded, but the uImage couldn't start.
The console hangs at:
Using FEC0 device
TFTP from server 10.193.100.158; our IP address is 10.193.102.233
Filename 'uImage_linaro'.
Load address: 0x70800000
Loading: #################################################################
#################################################################
#################################################################
#######################
done
Bytes transferred = 3193292 (30b9cc hex)
## Booting kernel from Legacy Image at 70800000 ...
Image Name: Linux-2.6.38-01026-gc9c8ead
Image Type: ARM Linux Kernel Image (uncompressed)
Data Size: 3193228 Bytes = 3 MB
Load Address: 10008000
Entry Point: 10008000
Verifying Checksum ... OK
Loading Kernel Image ... OK
OK
Starting kernel ...
Uncompressing Linux... done, booting the kernel.
Does anyone meet this before?
How can I fix it?
Thanks~~
Yours
Terry