[Cerowrt-devel] libmusl woes

[Cerowrt-devel] libmusl woes

[Dave Taht]

so I was very happy babel-1.6.1 and dnsmasq-2.73 landed, and wanted to
try the latest cake code, and maybe, finally deploy a couple routers
this week. I'd mostly just been working on debian/ubuntu x86 when I
had time to do anything technical...

...sometime in the past month openwrt switched to libmusl, a new
"cleaner" libc replacing uclibc. I am not sure how (or even if I
should) to switch back to uclibc, but it broke a half dozen+ apps I
used to build regularly: tcsh, libposix for lua, pimd, etherwake,
netperf, strongswan, xinetd, tinc, and I forget what else.

Of these I managed to build netperf, and just went down the list
disabling everything else. I happen to use tinc on the vpn, but I
don't need it right now...

And just as I finally got a new build, jonathon dropped a new version
of cake on me. I mean, not less than 5 seconds, after I got a build to
complete.

Attempting to build (for the first time) the mve stuff (linksys 1900,
1200ac) failed completely with a:

checking for arm-openwrt-linux-muslgnueabi-gcc...
/build/cero3/src/ac1900/build_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/gcc-linaro-4.8-2014.04-minimal/./gcc/xgcc
-B/build/cero3/src/ac1900/build_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/gcc-linaro-4.8-2014.04-minimal/./gcc/
-B/build/cero3/src/ac1900/staging_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/arm-openwrt-linux-muslgnueabi/bin/
-B/build/cero3/src/ac1900/staging_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/arm-openwrt-linux-muslgnueabi/lib/
-isystem /build/cero3/src/ac1900/staging_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/arm-openwrt-linux-muslgnueabi/include
-isystem /build/cero3/src/ac1900/staging_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/arm-openwrt-linux-muslgnueabi/sys-include
checking for suffix of object files... configure: error: in
`/build/cero3/src/ac1900/build_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/gcc-linaro-4.8-2014.04-minimal/arm-openwrt-linux-muslgnueabi/libgcc':
configure: error: cannot compute suffix of object files: cannot compile
See `config.log' for more details.
make[4]: *** [configure-target-libgcc] Error 1
make[4]: Leaving directory
`/build/cero3/src/ac1900/build_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/gcc-linaro-4.8-2014.04-minimal'


-- 
Dave Täht
What will it take to vastly improve wifi for everyone?
https://plus.google.com/u/0/explore/makewififast

Re: [Cerowrt-devel] libmusl woes

[Toke Høiland-Jørgensen]

Dave Taht <dave.taht@gmail.com> writes:

> I am not sure how (or even if I should) to switch back to uclibc,

Advanced configuration options (for developers) -> Toolchain Options ->
C library implementation. Trunk builds fine with uClibc as far as I can
tell...

Tried fixing tinc, but it appears to be a clash within the header files
with no solution forthcoming. See
http://www.openwall.com/lists/musl/2012/10/11/4

This script seems to fix it by modifying the Linux headers, but figure
that is not feasible for an openwrt package recipe...
https://github.com/lancethepants/tinc-arm-musl-static/blob/master/tinc.sh#L148

-Toke

Re: [Cerowrt-devel] libmusl woes

[Dave Taht]

On Wed, Jun 17, 2015 at 4:47 AM, Toke Høiland-Jørgensen <toke@toke.dk> wrote:
> Dave Taht <dave.taht@gmail.com> writes:
>
>> I am not sure how (or even if I should) to switch back to uclibc,

The bigger question is: is openwrt's future direction musl? I don't
mind switching, but would prefer to switch in the right direction.


> Advanced configuration options (for developers) -> Toolchain Options ->
> C library implementation. Trunk builds fine with uClibc as far as I can
> tell...

make dirclean helped clean up the mess. Going to try both uclibc and
musl build again. (really did want to "see" what an linksys 1900ac
did... did not anticipate it being so hard)


>
> Tried fixing tinc, but it appears to be a clash within the header files
> with no solution forthcoming. See
> http://www.openwall.com/lists/musl/2012/10/11/4
>
> This script seems to fix it by modifying the Linux headers, but figure
> that is not feasible for an openwrt package recipe...
> https://github.com/lancethepants/tinc-arm-musl-static/blob/master/tinc.sh#L148
>
> -Toke



-- 
Dave Täht
What will it take to vastly improve wifi for everyone?
https://plus.google.com/u/0/explore/makewififast

Re: [Cerowrt-devel] libmusl woes

[Dave Taht]

On Tue, Jun 16, 2015 at 11:05 PM, Dave Taht <dave.taht@gmail.com> wrote:
> so I was very happy babel-1.6.1 and dnsmasq-2.73 landed, and wanted to
> try the latest cake code, and maybe, finally deploy a couple routers
> this week. I'd mostly just been working on debian/ubuntu x86 when I
> had time to do anything technical...

https://github.com/dtaht/odhcp6c has something that works quite a bit better on
getting dhcp-pd to work without ever losing connectivity on debian.
(it is really awesome to see odhcp6c otherwise "just work" on debian)

Still, headaches:

https://github.com/sbyx/odhcp6c/issues/36

https://github.com/sbyx/odhcp6c/issues/35

because I am allergic still to hnetd, I just started reconstructing
what I used to do to get source specific ips "out there" for test.

https://github.com/dtaht/ipv6_selfassign

> ...sometime in the past month openwrt switched to libmusl, a new
> "cleaner" libc replacing uclibc. I am not sure how (or even if I
> should) to switch back to uclibc, but it broke a half dozen+ apps I
> used to build regularly: tcsh, libposix for lua, pimd, etherwake,
> netperf, strongswan, xinetd, tinc, and I forget what else.

A truly astonishing number of patches to fix compilation with musl
landed over the last few days.

xinetd, strongswan, tcsh, tinc fixes landed, among others.

It looks like openwrt has migrated to using pull requests for stuff. I
had emailed the fix for snmpd to jow, sent the one for netperf to
toke. I guess I will just start generating pull requests in the
future.

That leaves libposix and pimd and etherwake left busted in my default
stack of packages.

The box that I built with musl has been very stable, and also
sebastian landed a fix or three, and it looks like I need to update
tc-adv.

I also nagged rick jones to ship a netperf 2.7. The RTO behavior for
udp is important.

> Of these I managed to build netperf, and just went down the list
> disabling everything else. I happen to use tinc on the vpn, but I
> don't need it right now...
>
> And just as I finally got a new build, jonathon dropped a new version
> of cake on me. I mean, not less than 5 seconds, after I got a build to
> complete.
>
> Attempting to build (for the first time) the mve stuff (linksys 1900,
> 1200ac) failed completely with a:
>
> checking for arm-openwrt-linux-muslgnueabi-gcc...
> /build/cero3/src/ac1900/build_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/gcc-linaro-4.8-2014.04-minimal/./gcc/xgcc
> -B/build/cero3/src/ac1900/build_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/gcc-linaro-4.8-2014.04-minimal/./gcc/
> -B/build/cero3/src/ac1900/staging_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/arm-openwrt-linux-muslgnueabi/bin/
> -B/build/cero3/src/ac1900/staging_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/arm-openwrt-linux-muslgnueabi/lib/
> -isystem /build/cero3/src/ac1900/staging_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/arm-openwrt-linux-muslgnueabi/include
> -isystem /build/cero3/src/ac1900/staging_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/arm-openwrt-linux-muslgnueabi/sys-include
> checking for suffix of object files... configure: error: in
> `/build/cero3/src/ac1900/build_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/gcc-linaro-4.8-2014.04-minimal/arm-openwrt-linux-muslgnueabi/libgcc':
> configure: error: cannot compute suffix of object files: cannot compile
> See `config.log' for more details.
> make[4]: *** [configure-target-libgcc] Error 1
> make[4]: Leaving directory
> `/build/cero3/src/ac1900/build_dir/toolchain-arm_cortex-a9+vfpv3_gcc-4.8-linaro_musl-1.1.10_eabi/gcc-linaro-4.8-2014.04-minimal'

I am watching this build seemingly successfully now.

Wow. Night and day difference in headache.

>
> --
> Dave Täht
> What will it take to vastly improve wifi for everyone?
> https://plus.google.com/u/0/explore/makewififast



-- 
Dave Täht
worldwide bufferbloat report:
http://www.dslreports.com/speedtest/results/bufferbloat
And:
What will it take to vastly improve wifi for everyone?
https://plus.google.com/u/0/explore/makewififast