Hi,
* rewrote the Android.mk of libunwind to make use of autoreconf and libtool
* finished my work on libunwind
* upgraded my Linaro Android build environment
* debugged Linaro Android build failures (#891753)
* tested backtracing on the Linaro Android 2.3.5 and 2.3.7 branches
* documented debuggerd usage on Android:
https://wiki.linaro.org/Platform/Android/DebugAndroidSystemComponents
* Debugged a linking failure of the android perflab benchmark that Andy
is seeing. Turns out that the GCC is optimizing two consecutive calls to
sinf and cosf (same angle) are optimized by the GCC to one sincosf call.
The libm provided by the benchmark is lacking sincosf. Workaround is to
use -fno-builtin-sinf -fno-builtin-cosf.
Regards
Ken
== 64 bit atomics ==
* Still fighting membase
* Cleaned up a bunch of other issues, but I'm back at an 'expiry'
issue, where the test
stores some data with a fixed expiry time and then waits until after
it should have expired,
and checks it has. Except on ARM it sometimes doesn't expire quickly
enough. I've got
enough debug now to see that the server processes view of time (which
it updates via
an event about every second) is sometimes very behind gettimeofday()'s
view of time - and
have a small test for it. This doesn't seem to happen on x86. The
good part is that it's now
a much smaller test, the bad part is that it fails rarely - somewhere
between 1/1000 and 1/100
depending on its mood.
* Looked at a few other things to see if they might use 64 bit atomics:
- spice's (as in the VNC like protocol) FAQ said it needed 64bit
atomics and didn't work
on 32bit machines due to that; but the source appears to have been
fixed for 32bit.
- Looked at boost lock-free; it does have an implementation using
gcc's __sync primitives,
however for ARM it uses a hand coded set of primitives, those are
missing the 64 bit implementation,
but the contributor of the ARM code said that the boost lock-free
author preferred
not to use the gcc primtives.
== Other ==
* Testing latest libffi rc
- Had most of my varargs for hf fix in (had missed one part of a test)
* 1 day of non-linaro work
I'm on holiday next week.
Dave
(short week, 4 days)
RAG:
Red:
Amber: we still haven't settled on engineering blueprints and schedule
for the KVM work. Proceeding with some obviously necessary bits anyway
Amber: not clear whether we can do virtio-mmio this quarter
Green:
Current Milestones:
|| || Planned || Estimate || Actual ||
||upstream-omap3-cleanup || 2011-11-10 || 2011-11-10 || ||
(still no milestones, see above)
Historical Milestones:
||add-omap3-networking || 2011-10-13 || 2011-10-13 || 2011-10-13 ||
||a15-systemmode-planning || 2011-10-13 || 2011-10-13 || 2011-09-22 ||
||a15-usermode-support || 2011-11-10 || 2011-11-10 || 2011-10-27 ||
== upstream-omap3-cleanup ==
* spent a half day on this to try to get this blueprint tidied away
even if we aren't going to have much time for the later upstreaming
== other ==
* A15/KVM planning
* experimenting with getting an A15 kernel booting on Fast Model
(see https://wiki.linaro.org/LoicMinier/Sandbox/FastModels
and https://wiki.linaro.org/PeterMaydell/A15OnFastModels)
* estimated required work for doing a qemu model of a board for
one of the Landing Teams (ans: 6 man months +)
Hi,
- spent most of the week trying to reproduce regressions with vectorization
- started bringing the latest SLP feature, condition with different
types, to gcc-linaro. There are 5 patches. Merged one, started to
prepare another one.
- fixed PR 51112
Ira
Hi,
I've now put this at :
https://wiki.linaro.org/WorkingGroups/ToolChain/Meetings/2011-11-15
Are there any other topics that folks want to bring up ?
The one thing worth thinking about ahead of time is if we want to
bring ahead the call by an hour to allow Michael to join at a not so
crazy hour for him. What do folks think of 9 a.m. Tuesdays /
Wednesdays UTC ?
cheers
Ramana
Hey
When building u-boot for an ARMv5T platform (versatileqemu_config), the
Ubuntu-packaged Linaro cross-toolchain isn't suitable because it
only offers an ARMv7T2 libgcc. But I'd like the build to fail when
that happens rather than silently generating an u-boot.bin which will
trigger a SIGILL when it hits the first non-ARMv5T instruction.
I heard that gcc/ld are supposed to check this, but I'm not sure how
it's supposed to work; perhaps the way u-boot does its final link
prevents this from working properly?
I tried building u-boot as follows:
make O=obj-broken \
CROSS_COMPILE=arm-linux-gnueabi- ARCH=arm \
OPTFLAGS="-marm -march=armv5te" \
versatileqemu_config
make O=obj-broken \
CROSS_COMPILE=arm-linux-gnueabi- ARCH=arm \
OPTFLAGS="-marm -march=armv5te" \
-j2
The final link looks like this:
UNDEF_SYM=`arm-linux-gnueabi-objdump -x /home/lool/git/denx/u-boot/obj-v-broken/board/armltd/versatile/libversatile.o /home/lool/git/denx/u-boot/obj-v-broken/api/libapi.o /home/lool/git/denx/u-boot/obj-v-broken/arch/arm/cpu/arm926ejs/libarm926ejs.o /home/lool/git/denx/u-boot/obj-v-broken/arch/arm/cpu/arm926ejs/versatile/libversatile.o /home/lool/git/denx/u-boot/obj-v-broken/arch/arm/lib/libarm.o /home/lool/git/denx/u-boot/obj-v-broken/common/libcommon.o /home/lool/git/denx/u-boot/obj-v-broken/disk/libdisk.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/bios_emulator/libatibiosemu.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/block/libblock.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/dma/libdma.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/fpga/libfpga.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/gpio/libgpio.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/hwmon/libhwmon.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/i2c/libi2c.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/input/libinput.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/misc/libmisc.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/mmc/libmmc.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/mtd/libmtd.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/mtd/nand/libnand.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/mtd/onenand/libonenand.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/mtd/spi/libspi_flash.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/mtd/ubi/libubi.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/net/libnet.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/net/phy/libphy.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/pci/libpci.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/pcmcia/libpcmcia.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/power/libpower.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/rtc/librtc.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/serial/libserial.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/spi/libspi.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/twserial/libtws.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/usb/eth/libusb_eth.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/usb/gadget/libusb_gadget.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/usb/host/libusb_host.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/usb/musb/libusb_musb.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/usb/phy/libusb_phy.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/video/libvideo.o /home/lool/git/denx/u-boot/obj-v-broken/drivers/watchdog/libwatchdog.o /home/lool/git/denx/u-boot/obj-v-broken/fs/cramfs/libcramfs.o /home/lool/git/denx/u-boot/obj-v-broken/fs/ext2/libext2fs.o /home/lool/git/denx/u-boot/obj-v-broken/fs/fat/libfat.o /home/lool/git/denx/u-boot/obj-v-broken/fs/fdos/libfdos.o /home/lool/git/denx/u-boot/obj-v-broken/fs/jffs2/libjffs2.o /home/lool/git/denx/u-boot/obj-v-broken/fs/reiserfs/libreiserfs.o /home/lool/git/denx/u-boot/obj-v-broken/fs/ubifs/libubifs.o /home/lool/git/denx/u-boot/obj-v-broken/fs/yaffs2/libyaffs2.o /home/lool/git/denx/u-boot/obj-v-broken/lib/libfdt/libfdt.o /home/lool/git/denx/u-boot/obj-v-broken/lib/libgeneric.o /home/lool/git/denx/u-boot/obj-v-broken/lib/lzma/liblzma.o /home/lool/git/denx/u-boot/obj-v-broken/lib/lzo/liblzo.o /home/lool/git/denx/u-boot/obj-v-broken/lib/zlib/libz.o /home/lool/git/denx/u-boot/obj-v-broken/net/libnet.o /home/lool/git/denx/u-boot/obj-v-broken/post/libpost.o | sed -n -e 's/.*\(__u_boot_cmd_.*\)/-u\1/p'|sort|uniq`; cd /home/lool/git/denx/u-boot/obj-v-broken && arm-linux-gnueabi-ld -pie -T /home/lool/git/denx/u-boot/obj-v-broken/u-boot.lds -Bstatic -Ttext 0x10000 $UNDEF_SYM arch/arm/cpu/arm926ejs/start.o --start-group api/libapi.o arch/arm/cpu/arm926ejs/libarm926ejs.o arch/arm/cpu/arm926ejs/versatile/libversatile.o arch/arm/lib/libarm.o common/libcommon.o disk/libdisk.o drivers/bios_emulator/libatibiosemu.o drivers/block/libblock.o drivers/dma/libdma.o drivers/fpga/libfpga.o drivers/gpio/libgpio.o drivers/hwmon/libhwmon.o drivers/i2c/libi2c.o drivers/input/libinput.o drivers/misc/libmisc.o drivers/mmc/libmmc.o drivers/mtd/libmtd.o drivers/mtd/nand/libnand.o drivers/mtd/onenand/libonenand.o drivers/mtd/spi/libspi_flash.o drivers/mtd/ubi/libubi.o drivers/net/libnet.o drivers/net/phy/libphy.o drivers/pci/libpci.o drivers/pcmcia/libpcmcia.o drivers/power/libpower.o drivers/rtc/librtc.o drivers/serial/libserial.o drivers/spi/libspi.o drivers/twserial/libtws.o drivers/usb/eth/libusb_eth.o drivers/usb/gadget/libusb_gadget.o drivers/usb/host/libusb_host.o drivers/usb/musb/libusb_musb.o drivers/usb/phy/libusb_phy.o drivers/video/libvideo.o drivers/watchdog/libwatchdog.o fs/cramfs/libcramfs.o fs/ext2/libext2fs.o fs/fat/libfat.o fs/fdos/libfdos.o fs/jffs2/libjffs2.o fs/reiserfs/libreiserfs.o fs/ubifs/libubifs.o fs/yaffs2/libyaffs2.o lib/libfdt/libfdt.o lib/libgeneric.o lib/lzma/liblzma.o lib/lzo/liblzo.o lib/zlib/libz.o net/libnet.o post/libpost.o board/armltd/versatile/libversatile.o --end-group /home/lool/git/denx/u-boot/obj-v-broken/arch/arm/lib/eabi_compat.o -L /usr/lib/gcc/arm-linux-gnueabi/4.6.1 -lgcc -Map u-boot.map -o u-boot
I verified that all the .o files passed above have Tag_CPU_name:
"5TE" in their arm-linux-gnueabi-readelf -A output; the only
problematic file is -lgcc.
Note that the final link is done with arm-linux-gnueabi-ld and doesn't
set any architecture; I changed it manually to use gcc and pass the
-marm -march=armv5te, and had to set -nostdlib too when using gcc:
arm-linux-gnueabi-gcc -marm -march=armv5te -nostdlib -pie -T /home/lool/git/denx/u-boot/obj-v-broken/u-boot.lds -Bstatic -Ttext 0x10000 $UNDEF_SYM arch/arm/cpu/arm926ejs/start.o -Wl,--start-group api/libapi.o arch/arm/cpu/arm926ejs/libarm926ejs.o arch/arm/cpu/arm926ejs/versatile/libversatile.o arch/arm/lib/libarm.o common/libcommon.o disk/libdisk.o drivers/bios_emulator/libatibiosemu.o drivers/block/libblock.o drivers/dma/libdma.o drivers/fpga/libfpga.o drivers/gpio/libgpio.o drivers/hwmon/libhwmon.o drivers/i2c/libi2c.o drivers/input/libinput.o drivers/misc/libmisc.o drivers/mmc/libmmc.o drivers/mtd/libmtd.o drivers/mtd/nand/libnand.o drivers/mtd/onenand/libonenand.o drivers/mtd/spi/libspi_flash.o drivers/mtd/ubi/libubi.o drivers/net/libnet.o drivers/net/phy/libphy.o drivers/pci/libpci.o drivers/pcmcia/libpcmcia.o drivers/power/libpower.o drivers/rtc/librtc.o drivers/serial/libserial.o drivers/spi/libspi.o drivers/twserial/libtws.o drivers/usb/eth/libusb_eth.o drivers/usb/gadget/libusb_gadget.o drivers/usb/host/libusb_host.o drivers/usb/musb/libusb_musb.o drivers/usb/phy/libusb_phy.o drivers/video/libvideo.o drivers/watchdog/libwatchdog.o fs/cramfs/libcramfs.o fs/ext2/libext2fs.o fs/fat/libfat.o fs/fdos/libfdos.o fs/jffs2/libjffs2.o fs/reiserfs/libreiserfs.o fs/ubifs/libubifs.o fs/yaffs2/libyaffs2.o lib/libfdt/libfdt.o lib/libgeneric.o lib/lzma/liblzma.o lib/lzo/liblzo.o lib/zlib/libz.o net/libnet.o post/libpost.o board/armltd/versatile/libversatile.o -Wl,--end-group /home/lool/git/denx/u-boot/obj-v-broken/arch/arm/lib/eabi_compat.o -L /usr/lib/gcc/arm-linux-gnueabi/4.6.1 -lgcc -Wl,-Map u-boot.map -o u-boot
But this command works and produces an u-boot ELF which has
Tag_CPU_name: "7-A".
How would I break the build when libgcc isn't ARMv5T?
Thanks,
--
Loïc Minier
Spun Linaro GCC 4.6 release tarball, uploaded it to Michael's server,
and launched the testing.
Continued work on constant reuse optimization. I've now eliminated some
more false positives caused by inconsistent rtx_cost results. It turns
out the pass also fixes up inefficient constants generated by
arm_split_constants, which is nice.
Set yet more spec benchmark runs going as part of the generic tuning
investigation.
Other:
Half day Monday to recover from the weekend's travel.
Half day on internal Mentor activities.
== GDB ==
* Worked on support for cross-platform core file generation.
After some discussion on the mailing list it seems we've
come to an agreement that the remote protocol ought to have
two separate packets related to memory layout, one that
describes the permanent, system-wide layout (for embedded
systems) and one that describes the dynamic, per-process
layout (for processes with memory-mapped files). The latter
also ought to be integrated with the "info proc mappings"
command, which should work with gdbserver too.
I've been working on updating the patches accordingly.
== GCC ==
* Patch review week.
Mit freundlichen Gruessen / Best Regards
Ulrich Weigand
--
Dr. Ulrich Weigand | Phone: +49-7031/16-3727
STSM, GNU compiler and toolchain for Linux on System z and Cell/B.E.
IBM Deutschland Research & Development GmbH
Vorsitzender des Aufsichtsrats: Martin Jetter | Geschäftsführung: Dirk
Wittkopp
Sitz der Gesellschaft: Böblingen | Registergericht: Amtsgericht
Stuttgart, HRB 243294
Testing the SMS register pressure estimation on libav micro benchmarks
and eembc.
Discussed with Ayal the implementation. He had some ideas to consider
regarding the it.
Looking into the regressions of SMSed kernels in libav which are not
related to register pressure:
Consulting with Ayal regarding the case in dsputil-ssd_int8_vs_int16_c
where we have severe regression with SMS; it seemed that the
regression was due to dependence between accumulations that can be
avoided, more specifically we had the following case in vector code:
vec1 = vec1 + ...
...
vec1 = vec1+ ...
...
vec1 = vec1+ ...
...
vec1 = vec1+...
to resolve this, I implemented a hack similar to MVE optimiation in
the loop-unroller as follows:
vec1 = vec1 + ...
...
vec2 = vec2+ ...
...
vec3 = vec3+ ...
...
vec4 = vec4+...
This gives ~4.5% improvements to the non-SMSed version. The SMS
version now shows no regression as the problematic loop which caused
the regression now failed to be SMSed and I'm looking into the reason.
Another regression showed in idct-internal-8 is apparently related to
the do-loop optimziation (SMS actually failed to be applied in this
loop). when applying the patch to expand SMS to recognise doloop then
the regression is resolved.
(http://gcc.gnu.org/ml/gcc-patches/2011-09/msg02051.html; patch is not
in mainline yet)
