On 1/4/21 2:50 pm, scsijon wrote:
On 1/4/21 2:06 pm, scsijon wrote:
On 1/4/21 11:04 am, scsijon wrote:
On 30/3/21 5:04 pm, scsijon wrote:
I won't be communicating anything inline with this for a day or
two, i'd forgotten how slow one core is.
On 30/3/21 1:14 pm, scsijon wrote:
Since I now have a working 20-minimal packageset built, next is to
up it to 40-minimal-xorg.
First Error
Something doesn't like -j2, (2 cores) 3-binutils fails with
libtool errors, even though there is a work around for this
already configured in it. I have tried with a few other fixes, but
still fails so going to restart the whole build set as 1 core and
see what happens.
Ok, that didn't work except for a SLLLLLLLOOOOOOOOOWWWWWW build
until it stopped.
Started again back on 2 cores and >
0- with a bit of pre-read after the initial 40-~ fail rounds, I
decided to add a -j1 to libtool.desc as it hasn't a libtool.conf
(line 32 from binutils.conf with a [O] at the start of the line),
libtool builds anyway, but problems have been found with other
package builds sometimes if you don't, it seems to be the recomended
fix.
1- binutils build problem - A mod to the binutils.conf to # out the
--enable-shared line 28, it now builds and works for now, just means
binutils is static so I don't know what problems will appear for
later packages. There is quite a lot on the web about this problem,
it seems that it's relevant to all the 2.36.x series with various
work arounds, problem seems to be fixed with removing any of
"--enable-shared", or "-fno-plt" from CFLAGS, or "-Wl,-z,now" from
LDFLAGS. I've taken the first option as we had it in our
binutils.conf, which is what most people/linux-versions seem to have
done and recomend at this point of time. There is a bugzilla on this
one, 27360 (plus others linked), suppose to be fixed next version as
i read it.
2- xorg-video-intel is failing again, i'm yet to start to get to the
bottom of this one, had it in the past and don't remember the fix,
need to look up the past build logs and see what was done.
3- libevdev, now needs doxygen to build, i've added it in the
libevdev.cache and it built ok
4- nfs-utils, line needs adding in the nf-utils.conf to add a new
line 17 as>
var_append extraconfopt ' ' "--with-rpcgen"
apparently now needed if rpcgen isn't installed and 40-minimal-xorg
doesn't install the ?relevant package. Later we will have to make it
a conditional of course.
5- 5-linux is failing, something about /lib/udev but I wanted to get
some of the others working first so i didn't worry about it, we
don't use t2's linux package with the Puppy linux series anyway, but
I like it to build as a check of the overall build.
and that's the stage i'm up too,
time for breakfast/brunch and a sleep
scsijon
from 5-linux, a problem, it seems glibc isn't actually building
/usr/bin/rpcgen although there is a line in the glibc.conf to do so,
it's not in the current package's configure either to actually do
something and build it. Looks like we have a package hive-off or
something - under investigation
AH, looking at some of my old build notes, it early came to visabity
that rpcgen went from glibc to libtirpc and on to pcsvc-proto, where
it now resides, I have added that to my 40-minimal-xorg.in file, that
should sort that problem out, it also should allow me to revert
nfs-utils, but i'll test a build with it left in first. ?I do wonder
if it should be included and be built with one of the others though.
------------------------------------------------------
Well a full cleanup and fresh build started, lets see what fails this
time.
Ok, latest problems
1- Well somehow/somereason the develop/libtool package needs to be set
to 012-45---9, not just 01--45---9, the later causes a fail (just stops)
at 4-libtool, while with the extra 2 added it builds ok, don't know
whether it's something left over/missing from the rearangements Rene did
to the package stage coding or not, it just happens. Anyway, that's my
workaround for now.
2- mesa-demos has freeglut, a pre-requisite, in it's .cache file, but
for some reason it's not appearing with a X in the relevant
config/40/packages file, for now i've added it into my
40-minimal-xorg.in so it is set correctly and it builds.
3- rpcsvc-proto ocassionally fails to a rpcsvc-proto.err file, even
though it actually installs everything correctly, when it does fail it's
complaining about a duplicate directory, but it is not everytime, just
sometime, so i don't know why. I've set my configure to stop if a 5-
file fails, and all i'm doing is changing it from rpcsvc-proto.err to
rpcsvc-proto.log and it goes on in the build. Will need a fix somewhere
at a point in time, but I don't know/remember where or how Rene fixes these.
4- xf86-video-intel has gone back to failing for me, have no idea why,
error seems to complain about xorg-macros (xorg-util-macros in T2), even
though that's built and already installed. Maybe the xorg-macros is too
new for it as it's currently 19.1 and the xf86-video-intel is
complaining about wanting a minimum of 18.something. I've turned it
'off' with a O for now, my try installing an earlier 18.?
xorg-util-macros when i set my second T2 build base (hopefully in the
next week or two when another of my development boxes become free). Or
maybe a patch is needed for xf86-video-intel, I don't know.
5- By turning the stack smashing protector on in the Config Advanced
Settings, it causes 5-gcc to crash (badly) makes the build unuseable and
you need a ./scripts/Cleanup -full, turn it off in the config (-cfg 40
for me) and start again, seem to remember that the SSP hasmore than just
off and on nowadays, from memory I think it's now has 3 alternatives
'off', 'weak' and 'strong' . Something for rene when he has time to
investigate and add to his configure system.
6- 5-linux fails still, so i'm stopping here.
And now we restart with those fixes incorperated/added and well see what
happens next time.
---------------------------------------
(we are up to 5-gcc.out without a stop for a problem)
regards
scsijon
-----------------------------------------------------------
If you wish to unsubscribe from this mailing, send mail to
li...@t2-project.org with a subject of: unsubscribe t2
