Opened 4 years ago
Closed 4 years ago
#62063 closed defect (fixed)
mesa @19.0.8_0+osmesa+python27 patch failure on 10.6
Reported by: | rmottola (Riccardo) | Owned by: | jeremyhu (Jeremy Huddleston Sequoia) |
---|---|---|---|
Priority: | Normal | Milestone: | |
Component: | ports | Version: | |
Keywords: | patch snowleopard | Cc: | rlhamil, ballapete (Peter "Pete" Dyballa), macportsraf |
Port: | mesa |
Description
:info:patch Hunk #1 succeeded at 450 (offset -68 lines). :info:patch Command failed: cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_x11_mesa/mesa/work/mesa-19.0.8" && /usr/bin/patch -p1 < '/opt/local/var/macports/sources/rsync.macports.org/macports/release/tarballs/ports/x11/mesa/files/static-strndup.patch' :info:patch Exit code: 1 :error:patch Failed to patch mesa: command execution failed :debug:patch Error code: CHILDSTATUS 44780 1 :debug:patch Backtrace: command execution failed :
this is on 10.6 64bit
Attachments (1)
Change History (14)
Changed 4 years ago by rmottola (Riccardo)
comment:1 Changed 4 years ago by jmroot (Joshua Root)
Summary: | cannot build mesa - failed to patch → mesa @19.0.8_0+osmesa+python27 patch failure on 10.6 |
---|
comment:2 Changed 4 years ago by kencu (Ken)
I'm working on this one now -- just removing that patch, and it's building away now.
The legacysupport PG should allow more of mesa to "just work" without the patches -- but the d*mn libc++ / thread_local issue on 10.6.8 is a problem for the buildbot. I will go ahead soon and just override that I think.
comment:3 Changed 4 years ago by rlhamil
Cc: | rlhamil added |
---|
comment:4 Changed 4 years ago by ballapete (Peter "Pete" Dyballa)
Cc: | ballapete added |
---|
comment:5 follow-up: 9 Changed 4 years ago by ballapete (Peter "Pete" Dyballa)
On PPC Leopard, Mac OS X 10.5.8
I get a different patch failure:
MACOSX_DEPLOYMENT_TARGET='10.5' Executing: cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_nue.de.rsync.macports.org_macports_release_tarballs_ports_x11_mesa/mesa/work/mesa-19.0.8" && /usr/bin/patch -p1 < '/opt/local/var/macports/sources/nue.de.rsync.macports.org/macports/release/tarballs/ports/x11/mesa/files/no-missing-prototypes-error.patch' DEBUG: system: cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_nue.de.rsync.macports.org_macports_release_tarballs_ports_x11_mesa/mesa/work/mesa-19.0.8" && /usr/bin/patch -p1 < '/opt/local/var/macports/sources/nue.de.rsync.macports.org/macports/release/tarballs/ports/x11/mesa/files/no-missing-prototypes-error.patch' patching file configure.ac Hunk #1 FAILED at 292. 1 out of 1 hunk FAILED -- saving rejects to file configure.ac.rej Command failed: cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_nue.de.rsync.macports.org_macports_release_tarballs_ports_x11_mesa/mesa/work/mesa-19.0.8" && /usr/bin/patch -p1 < '/opt/local/var/macports/sources/nue.de.rsync.macports.org/macports/release/tarballs/ports/x11/mesa/files/no-missing-prototypes-error.patch' Exit code: 1
comment:6 Changed 4 years ago by kencu (Ken)
I fixed that patch failure trivially easily of course, but after that mesa doesn't build at present on 10.6 for other reasons (the OpenGL headers are different, and it won't compile).
So I just pegged it at 17.x on all the older systems <https://github.com/kencu/LeopardPorts/commit/3af9cdfb95e62503764f7bd99cdbb07a670db65e> while I decide when & how to fix that later.
comment:7 Changed 4 years ago by ballapete (Peter "Pete" Dyballa)
The file src/mesa/drivers/dri/common/xmlconfig.c
has gone, so static-strndup.patch could become smaller – but there are a few files that use strndup()
:
mesa-19.0.8/src/amd/common/ac_binary.c: binary->disasm_string = strndup(section_data->d_buf, mesa-19.0.8/src/compiler/spirv/spirv_to_nir.c: char *dup = ralloc_strndup(b, (char *)words, word_count * sizeof(*words)); mesa-19.0.8/src/egl/drivers/dri2/platform_x11.c: strndup(driver_name, mesa-19.0.8/src/egl/drivers/dri2/platform_x11.c: strndup(device_name, mesa-19.0.8/src/gallium/drivers/r600/evergreen_compute.c: binary->disasm_string = strndup(section_data->d_buf, mesa-19.0.8/src/gallium/drivers/radeonsi/si_debug.c: p = copy = strndup(p, semicolon - p); mesa-19.0.8/src/loader/loader.c: driver = strndup(version->name, version->name_len); mesa-19.0.8/src/util/ralloc.c:ralloc_strndup(const void *ctx, const char *str, size_t max) mesa-19.0.8/src/util/ralloc.h: * Like \c strndup, at most \p n characters are copied. If \p str is longer mesa-19.0.8/src/util/ralloc.h:char *ralloc_strndup(const void *ctx, const char *str, size_t n) MALLOCLIKE; mesa-19.0.8/src/util/strndup.h:strndup(const char *str, size_t max) mesa-19.0.8/src/util/xmlconfig.c: v->_string = strndup(string, STRING_CONF_MAXLEN);
The last one could be the new version of src/mesa/drivers/dri/common/xmlconfig.c
… (Anyway, I am going to wait for a solution.)
comment:8 Changed 4 years ago by macportsraf
Cc: | macportsraf added |
---|
comment:9 Changed 4 years ago by ballapete (Peter "Pete" Dyballa)
Replying to ballapete:
On
PPC Leopard, Mac OS X 10.5.8
I get a different patch failure:
On PPC Tiger, Mac OS X 10.4.11
, I see the same failure.
comment:10 follow-up: 11 Changed 4 years ago by ballapete (Peter "Pete" Dyballa)
There seems to be another bug, because port echo dependentof:mesa
reports:
cairo freeglut gtk3 libGLU
but port -vd upgrade libspectre
leads to the build failure of mesa
.
comment:11 Changed 4 years ago by ballapete (Peter "Pete" Dyballa)
Replying to ballapete:
There is no bug, there is another dependency
:
Library Dependencies: ghostscript, cairo
comment:12 Changed 4 years ago by kencu (Ken)
There has been a longstanding (since 2005) difference of opinion on one of mesa's variable definitions.
Various workarounds were put in place to handle the Apple differences.
For some reason that no longer works, at least on 10.6, and I just haven't dug in on why yet.
It has nothing to do with legacysupport, other ports, etc etc as far as I can see.
I'll get to it eventually. For now I pegged mesa as there are other pressing issues of higher priority, eg. a PR queue so long nobody could see the end of it, with dozens and dozens of DEAD PRs sitting around collection worms that needed to be cleared out.
comment:13 Changed 4 years ago by kencu (Ken)
Resolution: | → fixed |
---|---|
Status: | assigned → closed |
closed by [3ebe074a74ffe4e8880c733b5d4ef14b2a1c2ff6/macports-ports]
although this doesn't fix the build of mesa on < 10.7 yet...
full build log