From: Michael Heimpold <mhei@heimpold.de>
To: openembedded-devel@lists.openembedded.org
Cc: Khem Raj <raj.khem@gmail.com>
Subject: Re: [oe] [meta-oe][PATCH] php: drop explicite ARM_INSTRUCTION_SET
Date: Mon, 12 Jun 2023 19:57:51 +0200 [thread overview]
Message-ID: <3896518.ElGaqSPkdT@kerker> (raw)
In-Reply-To: <CAMKF1sqeRP-+YES+WoB784U1389_=f2MmuHzS13-d2rP2GK-Bw@mail.gmail.com>
Am Montag, 12. Juni 2023, 01:44:29 CEST schrieb Khem Raj:
> On Sun, Jun 11, 2023 at 1:53 PM Michael Heimpold <mhei@heimpold.de> wrote:
> > Am Sonntag, 11. Juni 2023, 18:32:31 CEST schrieb Khem Raj:
> > > On Sun, Jun 11, 2023 at 9:05 AM Michael Heimpold <mhei@heimpold.de>
> >
> > wrote:
> > > > If my git skills don't trick me, forcing ARM mode for PHP
> > > > dates back to year 2015 with commit e836f8f93.
> > > >
> > > > I wondered whether the compile problem still persists and just
> > > > tested that it compiles fine for qemuarm nowaydays.
> > > >
> > > > I also tested the binaries on a physical device, a
> > > > NXP iMX6ULL based one, and did not notice any problems
> > > > so far.
> > > >
> > > > Signed-off-by: Michael Heimpold <mhei@heimpold.de>
> > > > ---
> > > >
> > > > meta-oe/recipes-devtools/php/php_8.2.6.bb | 9 ---------
> > > > 1 file changed, 9 deletions(-)
> > > >
> > > > diff --git a/meta-oe/recipes-devtools/php/php_8.2.6.bb
> > > > b/meta-oe/recipes-devtools/php/php_8.2.6.bb index dad955452..404ff393b
> > > > 100644
> > > > --- a/meta-oe/recipes-devtools/php/php_8.2.6.bb
> > > > +++ b/meta-oe/recipes-devtools/php/php_8.2.6.bb
> > > > @@ -284,12 +284,3 @@ do_install:append:class-native() {
> > > >
> > > > create_wrapper ${D}${bindir}/php \
> > > >
> > > > PHP_PEAR_SYSCONF_DIR=${sysconfdir}/
> > > >
> > > > }
> > > >
> > > > -
> > > > -# Fails to build with thumb-1 (qemuarm)
> > > > -# | {standard input}: Assembler messages:
> > > > -# | {standard input}:3719: Error: selected processor does not support
> > > > Thumb mode `smull r0,r2,r9,r3' -# | {standard input}:3720: Error:
> > > > unshifted register required -- `sub r2,r2,r0,asr#31' -# | {standard
> > > > input}:3796: Error: selected processor does not support Thumb mode
> >
> > `smull
> >
> > > > r0,r2,r3,r3' -# | {standard input}:3797: Error: unshifted register
> > > > required -- `sub r2,r2,r0,asr#31' -# | make: ***
> > > > [ext/standard/math.lo]
> > > > Error 1
> > > > -ARM_INSTRUCTION_SET = "arm"
> > >
> > > Martin explained it well, it was seen with thumb1 ISA some armv6 and
> > > all armv7+ uses thumb2 ISA along with standard ARM ISA
> > > Testing this will need armv4/armv5 perhaps you can limit this to
> > > armv5/arm4 arches
> > >
> > > ARM_INSTRUCTION_SET:armv5 = "arm"
> > > ARM_INSTRUCTION_SET:armv4 = "arm"
> > >
> > > If you can test it on armv5 and ensure that it works there too now a
> > > days then we can remove it completely
> >
> > Thank you all for your feedback.
> >
> > Here is what I've done so far.
> >
> > I've done a compile test against machine qemuarmv5 which succeeded
> > without errors.
> >
> > I have a custom board at hand, which uses a NXP iMX28 CPU, the build
> > configuration tells me the same TUNE_FEATURES as qemuarmv5:
> > MACHINE = "evachargese"
> > TUNE_FEATURES = "arm armv5 thumb dsp"
> > I used the kirkstone branch to have a faster roundtrip since the board
> > currently runs a kirkstone-based system. The PHP binaries run without
> > obvious problems in a quick test (simple webpages on the device).
>
> That’s for testing it out. Can you check compiler longs for php and see if
> it passed -mthumb to it ?
Just to ensure, that I got you right: you mean something like
grep thumb tmp/work/armv5e-poky-linux-gnueabi/php/8.1.16-r0/temp/log.*
which shows an empty result for qemuarmv5/evachargese and for qemuarm
there are a lot of matches like:
.../log.do_compile.2431535:/bin/bash /.../cortexa15t2hf-neon-poky-linux-
gnueabi/php/8.1.16-r0/build/libtool --silent --preserve-dup-deps --tag CC --
mode=compile arm-poky-linux-gnueabi-gcc -mthumb -mfpu=neon -mfloat-abi=hard -
mcpu=cortex-a15 ...
> > I have no armv4 board at hand nor a found a configuration file for
> > qemuarmv4.
> > So I put some lines manually together using strongarm - which should be
> > armv4,
> > right? It resulted in
> > TUNE_FEATURES = "arm strongarm"
> > The build completed without error - no runtime testing with qemu done yet.
next prev parent reply other threads:[~2023-06-12 17:58 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-06-11 16:05 [meta-oe][PATCH] php: drop explicite ARM_INSTRUCTION_SET Michael Heimpold
2023-06-11 16:20 ` [oe] " Martin Jansa
2023-06-11 16:32 ` Khem Raj
2023-06-11 20:53 ` Michael Heimpold
2023-06-11 23:44 ` Khem Raj
2023-06-12 17:57 ` Michael Heimpold [this message]
2023-06-12 18:00 ` Khem Raj
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=3896518.ElGaqSPkdT@kerker \
--to=mhei@heimpold.de \
--cc=openembedded-devel@lists.openembedded.org \
--cc=raj.khem@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.