summaryrefslogtreecommitdiff
path: root/gnu/usr.bin/gcc
AgeCommit message (Collapse)Author
2009-07-22Fix the configure test for symbol visibility support; the current testMiod Vallat
tries to check for an old, known to be flawed, ld version, but the in-tree ld causes a false positive.
2009-06-24Correct the prototype of ffecom_gfrt_basictype() to match theJonathan Gray
function, this change was made in the 3.4.0 release of gcc. ok deraadt@
2009-06-06INTMAX_TYPE / UINTMAX_TYPE are always in terms of long long.Martynas Venckus
reported by sturm@. fixes libarchive on mips64. ok millert@, miod@. tested by jasper@.
2009-05-18INTMAX_TYPE / UINTMAX_TYPE are always in terms of long long.Martynas Venckus
reported by sturm@. fixes libarchive on sparc64 for naddy@, sturm@. mips is missing them too; but it has not been tested. ok millert@, miod@.
2009-04-17- enable libstdc++ pthread regressesKurt Miller
- fix g++ regress so it can find libstdc++ in the objdir okay espie@ kettenis@
2009-02-11fix gcc bug that broke double -> unsigned long conversions on sparc64Martynas Venckus
(TF floats between conversions). also fixes perl. reported and tested by naddy@. been in snaps for months. go ahead miod@
2008-11-11Add support for position independent executables. This adds compile flagsKurt Miller
-fpie & -fPIE which are similar to -fpic & -fPIC except that the resulting objects can not be used in shared libraries due to an additional optimization for pie. This also adds the -pie link flag which produces a pie executable given pic or pie objects. Combining -pie with -static is currently disabled. The only gcc3 arch that doesn't support pie at moment is arm due to binutils relocation bugs. ok kettenis@
2008-10-28Avoid infinite recursion in search_string_def() and search_pointer_def(),Miod Vallat
PR #5033; gcc 2.95 will need a similar fix (soon).
2008-09-29Add a -mno-ieee option to disable ieee mode; ok millert@Miod Vallat
2008-08-27exposed too much, only expose prototype outside ifdefMartynas Venckus
2008-08-27fix 128-bit division. gcc mangled arguments when passing to theMartynas Venckus
__udivti3, because MUST_PASS_IN_STACK always returned 1 on amd64; pr#5780 reported by Simon Kuhnle tested by Simon Kuhnle, sthen@, brad@ double-checked & tweak from miod@ ok sthen@, brad@
2008-06-09Make gcc on alpha use -mieee by default. This fixes many FPE problemsMichael Erdely
with several ports on alpha and does not produce a significant slowdown. Suggestion to patch gcc directly from millert@ Diff and ok from miod@ ok pvalchev@, martin@, deraadt@
2008-05-29Revert 1.6 and reenable -fno-regmove for now -- it stroke back in sha1(1).Miod Vallat
2008-05-08Due to the recent changes, it is no longer necessary to disable -fregmoveMiod Vallat
anymore.
2008-05-03add stmp-dirs to dependancy list, fixes a make -j issue seen on macppcDale Rahn
"looks okay" espie@
2008-04-25Prevent user asm statements from being put in delay slots, this avoidsMiod Vallat
warnings and/or unexpected operation when such statements are multi-insn macros, such as li/dli. Backported from gcc 3.3.6.
2008-04-25* PR rtl-optimization/13260Miod Vallat
``[3.4 Regression] Incorrect optimisation of loop termination condition'' This is a complete backport of this commit: 2003-12-04 J"orn Rennecke <joern.rennecke@superh.com> PR optimization/13260 * sh-protos.h (sh_expand_t_scc): Declare. * sh.h (PREDICATE_CODES): Add cmpsi_operand. * sh.c (cmpsi_operand, sh_expand_t_scc): New functions. * sh.md (cmpsi): Use cmpsi_operand. If T_REG is compared to something that is not a CONST_INT, copy it into a pseudo register. (subc): Fix description of new T value. (slt, sgt, sge, sgtu): Don't clobber T after rtl generation is over. (sltu, sleu, sgeu): Likewise. (seq, sne): Likewise. Use sh_expand_t_scc. * PR target/15886 ``[3.3/3.4/4.0 Regression] SH: Miscompilation with -O2 -fPIC'' This is a complete backport of this commit: 2004-06-11 J"orn Rennecke <joern.rennecke@superh.com> * sh.h (ALLOCATE_INITIAL_VALUE): Use return_address_pointer_rtx. * PR target/21255 ``%R and %S are not safe to use from asms'' This is a complete backport of this commit: 2005-08-31 J"orn Rennecke <joern.rennecke@st.com> PR target/21255 * sh.c (print_operand, %R and %S): Add handling of floating point registers, memory, constants and invalid operands. * Non PR - CAN_ELIMINATE() is modified to prevent fp -> sp elimination when a frame pointer needs to be preserved. I have serious reasons to think this will help propolice operation.
2008-04-25* PR target/11901Miod Vallat
``[3.3 only] set_fpscr isn't PIC even when __PIC__ defined; bad for prelinking, causes glibc-3.2.3 check-textrel test to fail'' Correctly address fpscr_values[] in set_fpscr() when compiled PIC. This is a partial backport of this commit: 2003-08-11 Kaz Kojima <kkojima@gcc.gnu.org> * config/sh/lib1funcs.asm (__udivdi3): Add .type and .size information in SHmedia case too. (__divdi3, __umoddi3, __moddi3, __init_trampoline, __ic_invalidate): Likewise. (__set_fpscr): Use an access via GOT for PIC case.
2008-04-25typoMiod Vallat
2008-04-25reg_renumber[] values are valid if >= 0, not > 0.Miod Vallat
2008-04-22I am so glad I'm a BSD developer.Miod Vallat
GNU developers wear grey. GNU developers use autoconf and friends to configure their software in the best possible way for the platform it is intended to run on. This frees us from the tedious job of, say, editing files hidden deep inside the sources to, say, tell the code that 64-bit platforms (you know, those which always define __LPA64__ and _LP64 among other things) have 64-bit integer types (one such platform even defines the completely unexpected symbol __mips64__.) GNU developers work much harder than we do, because they're so frightfully clever. I'm really awfully glad I'm a BSD developer.
2008-04-05don't force recreating directories in mklibgcc, this causes rebuilds duringMarc Espie
install. Prompted and tested by drahn@
2008-03-28Enable shared libraries on landisk.Dale Rahn
The sh architecture required some different linkage than most platforms, This fixes those other requirements.
2008-02-10give method name of covariant return issue.Marc Espie
okay miod@
2008-01-01In the alpha allocate_stack builtin, only trigger stack probes if not runningMiod Vallat
OpenBSD or VMS, or -fstack-check; this was missed when disabling stack checks a few months ago. ok kettenis@
2007-11-28add dependencies for protector.o, preventing it from being built too early.Marc Espie
Okay miod@
2007-11-26Add a new warning to gcc, -Wvariable-decl, which causes it to warn forMiod Vallat
auto declarations which size are not known at compile time. This flag will eventually be added to the kernel makefiles so that we can rely on -Wstack-larger-than work. ok deraadt@ mbalmer@ otto@ marco@
2007-11-20the extra stuff also needs stmp-dirs, as found out by naddy@Marc Espie
2007-10-29add missing dependency. Taken from upstream gcc.Marc Espie
okay kettenis@ (`if it builds' and it does)
2007-10-25fix dependency. ok millert@, \_o< miod@Marc Espie
2007-09-03Add __cxa_atexit() support for gcc3. This provides support for shared ↵Todd C. Miller
object destructors called at dlclose() time. Inspired by similar changes in FreeBSD and NetBSD.
2007-07-05On OpenBSD/hppa, not all space registers are equivalent, so don't setMark Kettenis
MASK_NO_SPACE_REGS. ok miod
2007-06-16INTMAX_TYPE and UINTMAX_TYPE are always in terms of long long onTodd C. Miller
OpenBSD, regardless of the size of a long. OK kettenis@ espie@
2007-05-31convert to new .Dd format;Jason McIntyre
2007-04-05Make sure all public declarations are declared .globl; remove code toMark Kettenis
declare functions .export, which is now redundant. This makes symbol aliases created with attribute("alias") visible. ok deraadt@
2007-03-18Disable stack frame checking in the prologue under OpenBSD, unless you compileMiod Vallat
with -fstack-check, for consistency with other platforms; ok kettenis@
2007-03-03Add new FUNC/ENDFUNC macros to generate size/type (function) for ELFDale Rahn
with suggestsion from kettenis. ok miod@
2007-03-02Move landisk to hardware floating point. At the moment the FPU context isMiod Vallat
always saved upon context switches, as FPU registers are heavily used for long long computations (don't ask). Gcc default to -m4. Credits to drahn@ otto@ and deraadt@ for feedback and help testing. Upgrade procedure if you don't want to use the damn snapshots: - build and install new kernel, reboot off it - build new gcc, do not install it yet - make includes - install new gcc - build and install lib/csu and lib/libc - make build
2007-01-20-Wstack-larger-than support, copied from gcc2.Miod Vallat
2007-01-17Remove references to xm files that no longer exist. Taken from GCC 3.4.Marc Aurele La France
ok brad@
2007-01-06Make gcc(1) really ignore __attribute__((visibility)) if HAVE_GAS_HIDDEN isn'tMark Kettenis
defined. ok miod@
2006-12-26Make exception handling work on alpha.Mark Kettenis
tested by naddy@, deanna@; ok brad@
2006-12-13Fix a few unitialized vars, all taken from gcc trunk, except theOtto Moerbeek
change in shl_sext_kind(). ok miod@ deraadt@
2006-11-28Turn OVERRIDE_OPTIONS macro into a function, so that we can putMiod Vallat
preprocessor directives in it; then do our changes only if either OPENBSD_NATIVE or OPENBSD_CROSS is defined; then disable -fregmove for now. ok otto@
2006-11-22Improved cross build support for gcc on OpenBSD. allows the cross builtDale Rahn
compiler to have the same default options as native. The changes to cross build a native compiler is not quite complete. "Get this in" miod@
2006-11-16hack at build process some more, so that files are not rebuilt under src,Marc Espie
especially since we have the tools. problem reported by Ryan McBride, okay deraadt@, mcbride@
2006-11-14Do not align jump targets to cache line boundaries at -O2 on OpenBSD.Miod Vallat
2006-11-11Provide size and type of symbols so that it doesn't confuse ld when buildingDale Rahn
shared binaries/libraries.
2006-11-11OpenBSD defaults to dynamic linker of /usr/libexec/ld.soDale Rahn
2006-11-10Yes, openbsd does not support shared libraries on sh yet, but why make it hard?Dale Rahn