| /[gentoo-x86]/eclass/toolchain-funcs.eclass |
Parent Directory
| 
Revision Log
| Links to HEAD: | (view) (download) (annotate) |
tc-arch: add more targets that are in at least linux-3.9
tc-arch-kernel: return arm64 for aarch64-* targets #449108 by James Cloos
reverted fatal error from unset and made it a warning only shown when checking for 'kern' arch, so that the portage environment call of tc-arch will not fail anymore, bug 436450
Fixed tc-ninja_magic_to_arch() to also use KV_FULL and fail if no kernel version specified, bug 432390
Remove redundant DESCRIPTION variable settings.
initial aarch64 support
also enable usr ldscripts for openbsd & netbsd #417451#38
enable generation of ldscripts for freebsd systems #428014 by Yuta SATOH
stop creating linker scripts for non darwin/linux systems #417451
unify build env var setup a bit so other packages can use it #391363
add *-softfp-* support to the vendor field for people who want this for binary compat with closed sourced drivers #416727 by Raúl Porcel; merge the tc-is-hardfloat helper into the one place where it is called in the whole tree to avoid further confusion between when the return values of tc-is-softfloat and tc-is-hardfloat
have tc-is-cross-compiler default to CHOST if CTARGET isnt set #414527 by Raúl Porcel
avoid multiple inclusions when possible to speed caching up
add econf_build helper for cross-compiling some packages
add more BUILD_xxx helpers, and unify the variable lookup logic a bit more in tc-getBUILD_PROG
quote flags value in export
unify tc-getPROG and tc-getBUILD_CC implementations, and add new tc-getBUILD_{CPP,CXX} helpers on top of that
output correct kernel directory for amd64-fbsd in tc-arch-kernel, bug #374893
tc-arch: re-indent the ppc code and check the $type before KV_to_int to avoid useless calculations
convert gcc version funcs to query the preprocessor for its gcc version to work with different compilers #335943 by Ambroz Bizjak
Make tc-getF77() fall back to the gcc-provided gfortran, not some nonexistent 'f77'.
add new tc-has-openmp helper func #320215 by Samuli Suominen
Add tc-is-hardfloat function, rename 'no' of tc-is-softfloat to 'target doesn't support softfloat', acked by vapier, bug #315987
tc-get{RC,DLLWRAP}: new helper functions for dealing with Windows toolchains
tc-has-tls: new thread local storage test function
start a tc-get helper for pkg-config
fix bug #283424 for real
Make the lib we're going to update using install_name_tool writable if it isn't. Restore write permissions afterwards, fixes bug #283424
add HPUX support for gen_usr_ldscript from Prefix
on Darwin we suffer from "funky sonames" as well (like ELF)
gen_usr_ldscript: * remove old USERLAND=Darwin hack * add install_name adjusting case for Darwin * add symlink creating cases for IRIX, AIX and HPUX * likewise for Interix and Windows * add support for -a for all platforms * add EPREFIX for Prefix compatability (could use inherit prefix for definition of EPREFIX variable) tc-is-static-only: * added function, currently only returning true for FreeMiNT (no shared libraries) reviewed by vapier
Revert 1.89 of toolchain-funcs.eclass, I should have reviewed it much better, and even worse it cases bugs like #264940
Add support for all Prefix arches, in particular for gen_usr_ld_script, and add AIX specific function, backport from Prefix
support funk SONAMEs like libfoo.so -> libfoo-0.1.so
add more comments to the linker script
Add tc-getOBJCOPY
first cut at a more automated method for ldscript installation
Add quoting to comparison on gcc-specs- funcs - fixes false results with GCC 4.x
Revert changes to hardened funcs for now
Knock off the version-fu and use test-flags-CC
Add exception for >=gcc-4 to gcc-specs-ssp-to-all func
Add some fno-strict-overflow stuff we missed. Thanks Magnus Granberg (Zorry).
tc-export: make sure the variable requested for export is a valid variable
[Fortran] Do a little reworking, since these functions remain unused by anything. Rename tc-getF90() to tc-getFC() for consistency with autotools, fortran.eclass, etc. Drop tc-getFORTRAN() -- people should use the appropriate function.
Fix tc-arch-kernel for freebsd: it is still i386 there, bug #222715
set locale to C when trying to parse the output of gcc as reported by Xake #217852
handle linux-2.6.24+ where {i386,x86_64}=>{x86}
add new documentation
filter unknown targets into the unknown category instead of falling back to the host arch #186073
add avr to the arch list
Resurrect tc-endian as I need it to control some testcases.
finish gen_usr_ldscript to toolchain-funcs from eutils
avoid eclass recursion
Upgrade gcc-specs-directive to handle spec string substitutions.
declare/shuffle around maintainers
add blackfin to softfloat-only case
first run at tc-is-softfloat()
no one uses tc-endian() so just cut it ... people should be handling endian in the source code itself anyways, not in the ebuild
allow gcc funcs to take an argument to pass down to tc-getCC #158239
fixup style in previous commit
The Darwin linker doesn't understand GNU ld scripts, instead it has no problems with symlinks.
move gen_usr_ldscript() to toolchain-funcs.eclass and make it smarter
Add tc-getSTRIP. With gcc-4.X, GNU fortran is no longer f77, so add tc-getF90, and tc-getFORTRAN.
Add function to get CPP.
Make gcc-specs-directive work on non-English locales, bug #127742
Apply whitespace checks from repoman to eclasses.
Minor fix for the previous patch to keep ppc64 working.
Applied fix for bug #119203.
oops, had blackfin args reversed
add support for embedded arches blackfin/nios2
Minor fixup for new powerpc tree in toolchain-funcs.eclass.
remove sep case for powerpc64 hosts and fixup the unified code
add support for newer ppc kernels thanks to hansmi
*** empty log message ***
Get rid of ABI crap since it's needed only in dev profiles, and with gcc-config-2.0 those profiles can start using CHOST rather than ABI to properly set the toolchain.
sort arches by alphabetical order and define a few more
include dumpspecs output as first data for specs parsing parse additional-style specs as well as replacement specs add function to detect auto-upgrade of ssp to ssp-all
make sure tc-getBUILD_CC() properly falls back to gcc #74101
add arm thumb support
handle CBUILD better since it is forced in portage now and check for CC_FOR_BUILD BUILD_CC HOSTCC as the build cc
Fix borkage with gcc4.
Added functions to read/check gcc specs (for use in checking hardened toolchain state)
QA: Apply whitespace checks from repoman to eclasses.
Remove ECLASS= and INHERITED= which haven't been necessary for a long time now
Remove ECLASS= and INHERITED= which haven't been necessary for a long time now
Changes for ppc64 64bit kernel / 32bit userland
dont echo variables when exporting anymore
Fallback to CHOST if $(get_abi_CHOST)-gcc doesn't exist.
Add support for using get_abi_CHOST as a prefix before CHOST.
undo style changes
Added sanity checks about for threads and shared with EXTRA_ECONF. Added support for CTARGET_ALIASES with >=sys-devel/gcc-config-1.4.0.
make endian func a little more robust and change sh behavior to match binutils (default is little)
Support - in the gcc version string
make tc-endian act like tc-arch wrt to default CHOST/CTARGET selection
make fallbak tc-arch() $ARCH
Correct ARCH setting on sparc64.
touchup copyright/whitespace
Fixed bad ninja.
tc-arch: default to $CTARGET and then fall back to $CHOST
add m68k to our list
kernel arch for x86 is i386, oops
dont need eutils eclass
change the names of the arch functions to be like all the other tc-funcs. also add a func to figure out endian.
add function to translate CHOST tuples into ARCH for portage/kernel
dont kill $prog unless $search is valid
style update and remove $CTARGET logic from tc-get()
force cross-compiling users to set CBUILD
use CBUILD / CHOST in the default case of tc-is-cross-compiler
err need to negate the logic
make sure we return the correct status
tc-is-cross-compiler()
fix how BUILD_CC is defined (you know, so it works :D)
tc-export()
add a bit more logic to buildcc()
tc-getGCJ
add fortran wrapper tc-getF77
favor $CTARGET over $CHOST
oops, had inverse logic #68106
tc-getBUILD_CC()
fix $ECLASS
mmm, undo the renaming of gcc version stuff
generic gcc successor
This form allows you to request diffs between any two revisions of this file. For each of the two "sides" of the diff, select a symbolic revision name using the selection box, or choose 'Use Text Field' and enter a numeric revision.
| ViewVC Help | |
| Powered by ViewVC 1.1.13 |