From ca34d7a7c77fa8dc823fd05439870e5a96610c5a Mon Sep 17 00:00:00 2001 From: Ulrich Drepper Date: Thu, 14 Aug 1997 01:54:13 +0000 Subject: Update. 1997-08-14 03:14 Ulrich Drepper * elf/dl-minimal.c: Don't use relative #include paths. (malloc): Don't try to initialize _dl_pagesize, it already is. (__sigjmp_save, longjmp): Reformat. * sysdeps/generic/dl-sysdep.c: Implement _dl_show_auxv function to show content of auxiliary array. (_dl_sysdep_start): Remember start of auxiliary vector. * sysdeps/mach/hurd/dl-sysdep.c: (_dl_show_auxv): New dummy function. * elf/link.h: Add prototype for _dl_show_auxv. * elf/rtld.c (dl_main): Call _dl_show_auxv if LD_SHOW_AUXV envvar is present. * libio/iofdopen.c: Better prepare for use in libstdc++. * libio/iofflush.c: Likewise. * libio/iofgetpos.c: Likewise. * libio/iofgets.c: Likewise. * libio/iofopen.c: Likewise. * libio/iofputs.c: Likewise. * libio/iofread.c: Likewise. * libio/iofsetpos.c: Likewise. * libio/ioftell.c: Likewise. * libio/iofwrite.c: Likewise. * libio/iogetdelim.c: Likewise. * libio/iogets.c: Likewise. * libio/iopopen.c: Likewise. * libio/ioputs.c: Likewise. * libio/iosetbuffer.c: Likewise. * libio/iosetvbuf.c: Likewise. * libio/ioungetc.c: Likewise. * libio/iovsprintf.c: Likewise. * libio/iovsscanf.c: Likewise. * manual/creature.texi: Mention _XOPEN_SOURCE_EXTENDED in correct place. * manual/math.texi: Add comment to sincos decsription to say it's a GNU extension. Fix description of random function. * posix/fnmatch.h: Recognize Windog by _WINDOWS32 not WIN32. * posix/glob.c: Likewise. Handle stupid system headers on SunOS. Add casts for all __alloca calls. * posix/glob.h: Recognize Windog by _WINDOWS32 not WIN32. Declare additional GLOB_* constants also if _GNU_SOURCE is defined. Patches by Paul D. Smith . * stdio-common/_itoa.c: Don't use relateive #include paths. * sysdeps/i386/fpu/bits/mathinline.h: Enable optimizations even for gcc 2.7. (atan): Loading 1.0 from memory does not pay off. (asinh): Fix typo. 1997-08-13 09:44 Philip Blundell * sysdeps/stub/accept.c: Use socklen_t where necessary. * sysdeps/stub/bind.c: Likewise. * sysdeps/stub/connect.c: Likewise. * sysdeps/stub/getsockname.c: Likewise. * sysdeps/stub/getsockopt.c: Likewise. * sysdeps/stub/recvfrom.c: Likewise. * sysdeps/stub/sendto.c: Likewise. * sysdeps/stub/setsockopt.c: Likewise. 1997-08-11 Andreas Schwab * io/ftw.c (object_compare): Avoid memcmp when comparing scalar types. 1997-08-12 Andreas Schwab * math/libm-test.c (cbrt_test): Add epsilon for long double to `cbrt (-0.001)'. 1997-08-11 Andreas Schwab * sysdeps/stub/bits/string.h: Fix comment. 1997-08-01 Andreas Schwab * elf/Makefile (routines): Remove dl-version. (dl-routines): Add it here instead. This removes dead code from libc.so. 1997-08-12 17:17 Wolfram Gloger * malloc/thread-m.h (thread_atfork, thread_atfork_static): New macros to encapsulate pthread_atfork's functionality. * malloc/malloc.c (ptmalloc_lock_all, ptmalloc_unlock_all): New functions, registered via thread_atfork in the initialization routine or via thread_atfork_static in global scope. This ensures a consistent state of all locks across fork(). 1997-08-11 23:00:32 1997 Mark Kettenis * login/programs/xtmp.c (compare_entry): Partially undo patch made by Klaus Espenlaub. There was no xtmp/utmp typo. Added comment to cclarify this. * login/programs/xtmp.h: Likewise. * posix/getopt.c: Recognize Windog by _WINDOWS32 not WIN32. Patch by Paul D. Smith . 1997-08-12 13:51 Ulrich Drepper * login/getutid.c: Rename to __getutid and make getutid and getutxid weak aliases. Patch by ir. Mark M._Kettenis . 1997-08-11 23:55 Ulrich Drepper * sysdeps/unix/sysv/linux/Makefile [subdir=misc] (sysdep_headers): Add bits/mman.h. 1997-08-11 08:51:21 1997 H.J. Lu * stdlib/fmtmsg.c (fmtmsg): Add static to __libc_once_define macro. * libc.map (__progname_full, __progname): Added. 1997-08-04 08:31 H.J. Lu * sysdeps/unix/sysv/linux/if_index.c (if_nametoindex): Return 0 if SIOGIFINDEX is not defined. (if_indextoname): Initialize "result" to NULL. (if_nametoindex, if_nameindex): Fix ENOSYS setting. 1997-08-11 11:58 Richard Henderson Sparc ISO C 9x exception handling: * sysdeps/sparc/fpu/bits/mathdef.h: New file. * sysdeps/sparc/fpu/fclrexcpt.c: New file. * sysdeps/sparc/fpu/fegetenv.c: New file. * sysdeps/sparc/fpu/fegetround.c: New file. * sysdeps/sparc/fpu/feholdexcpt.c: New file. * sysdeps/sparc/fpu/fesetenv.c: New file. * sysdeps/sparc/fpu/fesetround.c: New file. * sysdeps/sparc/fpu/feupdateenv.c: New file. * sysdeps/sparc/fpu/fgetexcptflg.c: New file. * sysdeps/sparc/fpu/fraiseexcpt.c: New file. * sysdeps/sparc/fpu/fsetexcptflg.c: New file. * sysdeps/sparc/fpu/ftestexcept.c: New file. * sysdeps/sparc/sparc32/fpu/bits/fenv.h: New file. * sysdeps/sparc/sparc64/fpu/bits/fenv.h: New file. 1997-08-11 11:58 Richard Henderson Sparc64 merge: * configure.in: Change up subdirectories, rename sparc8->sparcv8. * csu/Makefile: Nuke need-nopic-initfini bits. The problem was that sparc64 ld did not recognize pic relocs at all. * elf/dl-deps.c: Include . * elf/do-rel.h (elf_dynamic_do_rel): Take reloc address and size directly instead of the DT names to get them from. * elf/dynamic-link.h (_ELF_DYNAMIC_DO_RELOC): New macro distilled from ELF_DYNAMIC_DO_REL{,A}. Define a second version that accounts for the possibility that DT_RELA and DT_JMPPLT overlap, as such is the case on the Sparc. * elf/elf.h: Stub out Elf64_Ver*, and make them the same as the Elf32_Ver* bits. The linker does not even know to differentiate between the word sizes. Someone should examine this and figure out what the proper sizes of things should be for 64-bit hosts before the sparc64 libraries are distributed. * shlib-versions: Rename sparc64 ld.so to not conflict with sparc32 on the same machine. Move old sysdeps/sparc contents to sysdeps/sparc/sparc32, rename sparc8 subdir to sparcv8, move sysdeps/sparc64 content to sysdeps/sparc/sparc64. Same for sysdeps/unix/sysv/linux/sparc. * sysdeps/sparc/Implies: Do ieee754 here. * sysdeps/sparc/configure.in: Move from sparc32/. * sysdeps/sparc/sparc32/Implies: Do wordsize here. * sysdeps/sparc/sparc64/Implies: Likewise. * sysdeps/sparc/sparc32/dl-machine.h (elf_machine_load_address): Use uninitialized .got to find load address instead of AT_BASE, as the later does not exist when running ld.so directly. * sysdeps/sparc/sparc32/fpu/fpu_control.h: Move from ../. * sysdeps/sparc/sparc64/fpu/fpu_control.h: Likewise. * sysdeps/sparc/sparc32/udiv_qrnnd.S: Clean up PIC support. * sysdeps/sparc/sparc64/Makefile (sysdep_routines): Add s_finitel and s_signbitl for stdio. * sysdeps/sparc/sparc64/configure.in: Remove, we no longer need nopic_initfini. * sysdeps/sparc/sparc64/bsd-_setjmp.S: Remove. * sysdeps/sparc/sparc64/bsd-setjmp.S: Remove. * sysdeps/unix/sysv/linux/sparc/sparc64/bsd-_setjmp.S: New empty file. * sysdeps/unix/sysv/linux/sparc/sparc64/bsd-setjmp.S: New empty file. * sysdeps/unix/sysv/linux/sparc/sparc64/longjmp.S: Add __longjmp alias. * sysdeps/unix/sysv/linux/sparc/sparc64/setjmp.S: Add weak _setjmp and setjmp entry points. * sysdeps/sparc/sparc64/dl-machine.h: Finish real implementation. * sysdeps/sparc/sparc64/elf/crtbegin.S: Add copyright, fix PIC. * sysdeps/sparc/sparc64/elf/crtend.S: Likewise. * sysdeps/unix/sysv/linux/sparc/bits/sigaction.h (struct sigaction): Fix sa_flags size for sparc64; add sa_restorer for kernel. * sysdeps/unix/sysv/linux/sparc/sparc64/brk.S: Fix PIC. * sysdeps/unix/sysv/linux/sparc/sparc64/clone.S: Likewise. * sysdeps/unix/sysv/linux/sparc/sparc64/init-first.h: Likewise. * sysdeps/unix/sysv/linux/sparc/sparc64/sysdep.h: Likewise. * sysdeps/unix/sysv/linux/sparc/sparc64/syscall.S: New file. 1997-08-10 18:29:08 1997 Mark Kettenis * login/login.c (login): Check for correct return value of utmpname. 1997-08-11 16:49 Ulrich Drepper * time/tzset.c (__tzset_internal): Correctly handle TZ strings following incorrect old POSIX specs. Patch by Paul Eggert . * sysdeps/generic/bits/sigaction.h: Correct comment for SA_RESTART. * sysdeps/unix/bsd/osf/bits/sigaction.h: Likewise. * sysdeps/unix/sysv/linux/alpha/bits/sigaction.h: Likewise. * sysdeps/unix/sysv/linux/bits/sigaction.h: Likewise. * sysdeps/unix/sysv/linux/mips/bits/sigaction.h: Likewise. * sysdeps/unix/sysv/linux/sparc/bits/sigaction.h: Likewise. * sysdeps/unix/sysv/minix/bits/sigaction.h: Likewise. * sysdeps/unix/sysv/sysv4/bits/sigaction.h: Likewise. Reported by NIIBE Yutaka . 1997-08-09 Andreas Schwab * manual/math.texi: Define the macros @mul and @infinity and use them to format the multiplication and infinity signs. (FP Comparison Functions): Use @code, not @math, for C code examples. * elf/dl-profile.c: Add descriptions and implement reading of --- sysdeps/sparc/sparc32/sdiv.S | 369 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 369 insertions(+) create mode 100644 sysdeps/sparc/sparc32/sdiv.S (limited to 'sysdeps/sparc/sparc32/sdiv.S') diff --git a/sysdeps/sparc/sparc32/sdiv.S b/sysdeps/sparc/sparc32/sdiv.S new file mode 100644 index 0000000000..02ed2e973c --- /dev/null +++ b/sysdeps/sparc/sparc32/sdiv.S @@ -0,0 +1,369 @@ + /* This file is generated from divrem.m4; DO NOT EDIT! */ +/* + * Division and remainder, from Appendix E of the Sparc Version 8 + * Architecture Manual, with fixes from Gordon Irlam. + */ + +/* + * Input: dividend and divisor in %o0 and %o1 respectively. + * + * m4 parameters: + * .div name of function to generate + * div div=div => %o0 / %o1; div=rem => %o0 % %o1 + * true true=true => signed; true=false => unsigned + * + * Algorithm parameters: + * N how many bits per iteration we try to get (4) + * WORDSIZE total number of bits (32) + * + * Derived constants: + * TOPBITS number of bits in the top decade of a number + * + * Important variables: + * Q the partial quotient under development (initially 0) + * R the remainder so far, initially the dividend + * ITER number of main division loop iterations required; + * equal to ceil(log2(quotient) / N). Note that this + * is the log base (2^N) of the quotient. + * V the current comparand, initially divisor*2^(ITER*N-1) + * + * Cost: + * Current estimate for non-large dividend is + * ceil(log2(quotient) / N) * (10 + 7N/2) + C + * A large dividend is one greater than 2^(31-TOPBITS) and takes a + * different path, as the upper bits of the quotient must be developed + * one bit at a time. + */ + + + +#include "sysdep.h" +#ifdef __linux__ +#include +#else +#ifdef __svr4__ +#include +#else +#include +#endif +#endif + +ENTRY(.div) + ! compute sign of result; if neither is negative, no problem + orcc %o1, %o0, %g0 ! either negative? + bge 2f ! no, go do the divide + xor %o1, %o0, %g6 ! compute sign in any case + tst %o1 + bge 1f + tst %o0 + ! %o1 is definitely negative; %o0 might also be negative + bge 2f ! if %o0 not negative... + sub %g0, %o1, %o1 ! in any case, make %o1 nonneg +1: ! %o0 is negative, %o1 is nonnegative + sub %g0, %o0, %o0 ! make %o0 nonnegative +2: + + ! Ready to divide. Compute size of quotient; scale comparand. + orcc %o1, %g0, %o5 + bne 1f + mov %o0, %o3 + + ! Divide by zero trap. If it returns, return 0 (about as + ! wrong as possible, but that is what SunOS does...). + ta ST_DIV0 + retl + clr %o0 + +1: + cmp %o3, %o5 ! if %o1 exceeds %o0, done + blu Lgot_result ! (and algorithm fails otherwise) + clr %o2 + sethi %hi(1 << (32 - 4 - 1)), %g1 + cmp %o3, %g1 + blu Lnot_really_big + clr %o4 + + ! Here the dividend is >= 2**(31-N) or so. We must be careful here, + ! as our usual N-at-a-shot divide step will cause overflow and havoc. + ! The number of bits in the result here is N*ITER+SC, where SC <= N. + ! Compute ITER in an unorthodox manner: know we need to shift V into + ! the top decade: so do not even bother to compare to R. + 1: + cmp %o5, %g1 + bgeu 3f + mov 1, %g7 + sll %o5, 4, %o5 + b 1b + add %o4, 1, %o4 + + ! Now compute %g7. + 2: addcc %o5, %o5, %o5 + bcc Lnot_too_big + add %g7, 1, %g7 + + ! We get here if the %o1 overflowed while shifting. + ! This means that %o3 has the high-order bit set. + ! Restore %o5 and subtract from %o3. + sll %g1, 4, %g1 ! high order bit + srl %o5, 1, %o5 ! rest of %o5 + add %o5, %g1, %o5 + b Ldo_single_div + sub %g7, 1, %g7 + + Lnot_too_big: + 3: cmp %o5, %o3 + blu 2b + nop + be Ldo_single_div + nop + /* NB: these are commented out in the V8-Sparc manual as well */ + /* (I do not understand this) */ + ! %o5 > %o3: went too far: back up 1 step + ! srl %o5, 1, %o5 + ! dec %g7 + ! do single-bit divide steps + ! + ! We have to be careful here. We know that %o3 >= %o5, so we can do the + ! first divide step without thinking. BUT, the others are conditional, + ! and are only done if %o3 >= 0. Because both %o3 and %o5 may have the high- + ! order bit set in the first step, just falling into the regular + ! division loop will mess up the first time around. + ! So we unroll slightly... + Ldo_single_div: + subcc %g7, 1, %g7 + bl Lend_regular_divide + nop + sub %o3, %o5, %o3 + mov 1, %o2 + b Lend_single_divloop + nop + Lsingle_divloop: + sll %o2, 1, %o2 + bl 1f + srl %o5, 1, %o5 + ! %o3 >= 0 + sub %o3, %o5, %o3 + b 2f + add %o2, 1, %o2 + 1: ! %o3 < 0 + add %o3, %o5, %o3 + sub %o2, 1, %o2 + 2: + Lend_single_divloop: + subcc %g7, 1, %g7 + bge Lsingle_divloop + tst %o3 + b,a Lend_regular_divide + +Lnot_really_big: +1: + sll %o5, 4, %o5 + cmp %o5, %o3 + bleu 1b + addcc %o4, 1, %o4 + be Lgot_result + sub %o4, 1, %o4 + + tst %o3 ! set up for initial iteration +Ldivloop: + sll %o2, 4, %o2 + ! depth 1, accumulated bits 0 + bl L.1.16 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + ! depth 2, accumulated bits 1 + bl L.2.17 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + ! depth 3, accumulated bits 3 + bl L.3.19 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + ! depth 4, accumulated bits 7 + bl L.4.23 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + b 9f + add %o2, (7*2+1), %o2 + +L.4.23: + ! remainder is negative + addcc %o3,%o5,%o3 + b 9f + add %o2, (7*2-1), %o2 + + +L.3.19: + ! remainder is negative + addcc %o3,%o5,%o3 + ! depth 4, accumulated bits 5 + bl L.4.21 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + b 9f + add %o2, (5*2+1), %o2 + +L.4.21: + ! remainder is negative + addcc %o3,%o5,%o3 + b 9f + add %o2, (5*2-1), %o2 + + + +L.2.17: + ! remainder is negative + addcc %o3,%o5,%o3 + ! depth 3, accumulated bits 1 + bl L.3.17 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + ! depth 4, accumulated bits 3 + bl L.4.19 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + b 9f + add %o2, (3*2+1), %o2 + +L.4.19: + ! remainder is negative + addcc %o3,%o5,%o3 + b 9f + add %o2, (3*2-1), %o2 + + +L.3.17: + ! remainder is negative + addcc %o3,%o5,%o3 + ! depth 4, accumulated bits 1 + bl L.4.17 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + b 9f + add %o2, (1*2+1), %o2 + +L.4.17: + ! remainder is negative + addcc %o3,%o5,%o3 + b 9f + add %o2, (1*2-1), %o2 + + + + +L.1.16: + ! remainder is negative + addcc %o3,%o5,%o3 + ! depth 2, accumulated bits -1 + bl L.2.15 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + ! depth 3, accumulated bits -1 + bl L.3.15 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + ! depth 4, accumulated bits -1 + bl L.4.15 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + b 9f + add %o2, (-1*2+1), %o2 + +L.4.15: + ! remainder is negative + addcc %o3,%o5,%o3 + b 9f + add %o2, (-1*2-1), %o2 + + +L.3.15: + ! remainder is negative + addcc %o3,%o5,%o3 + ! depth 4, accumulated bits -3 + bl L.4.13 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + b 9f + add %o2, (-3*2+1), %o2 + +L.4.13: + ! remainder is negative + addcc %o3,%o5,%o3 + b 9f + add %o2, (-3*2-1), %o2 + + + +L.2.15: + ! remainder is negative + addcc %o3,%o5,%o3 + ! depth 3, accumulated bits -3 + bl L.3.13 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + ! depth 4, accumulated bits -5 + bl L.4.11 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + b 9f + add %o2, (-5*2+1), %o2 + +L.4.11: + ! remainder is negative + addcc %o3,%o5,%o3 + b 9f + add %o2, (-5*2-1), %o2 + + +L.3.13: + ! remainder is negative + addcc %o3,%o5,%o3 + ! depth 4, accumulated bits -7 + bl L.4.9 + srl %o5,1,%o5 + ! remainder is positive + subcc %o3,%o5,%o3 + b 9f + add %o2, (-7*2+1), %o2 + +L.4.9: + ! remainder is negative + addcc %o3,%o5,%o3 + b 9f + add %o2, (-7*2-1), %o2 + + + + + 9: +Lend_regular_divide: + subcc %o4, 1, %o4 + bge Ldivloop + tst %o3 + bl,a Lgot_result + ! non-restoring fixup here (one instruction only!) + sub %o2, 1, %o2 + + +Lgot_result: + ! check to see if answer should be < 0 + tst %g6 + bl,a 1f + sub %g0, %o2, %o2 +1: + retl + mov %o2, %o0 -- cgit 1.4.1