HACKS revision 1.22
11.22Smrg# $NetBSD: HACKS,v 1.22 2003/11/05 00:50:05 mrg Exp $ 21.1Slukem# 31.1Slukem# This file is intended to document workarounds for currently unsolved 41.1Slukem# (mostly) compiler bugs. 51.1Slukem# 61.1Slukem# Format: 71.1Slukem# hack title 81.1Slukem# cdate creation date 91.1Slukem# mdate mod date 101.1Slukem# who responsible developer 111.1Slukem# port ... 121.1Slukem# affected ports, space separated, if not "all" 131.1Slukem# file affected file : revision : line from : line to 141.1Slukem# affected files and revision and line numbers describing hack 151.1Slukem# multiple lines if necessary. 161.1Slukem# pr NNNN ... 171.1Slukem# problem reports this hack works around, if known. Space 181.1Slukem# separated. 191.1Slukem# regress src/regress/directory/where/test/found 201.1Slukem# regression test directories, if available. 211.1Slukem# descr 221.1Slukem# insert short informal description (multi-line). (Longer ones 231.1Slukem# should be in the PR database. More formal descriptions might 241.1Slukem# be in the regress tree. See above). 251.1Slukem# kcah 261.1Slukem# closing bracket. 271.1Slukem# 281.1Slukem# this is a comment. 291.1Slukem 301.1Slukemhack netstat ieee1394 address printing. 311.1Slukemmdate 14 Nov 2000 321.1Slukemwho matt 331.1Slukemfile lib/libc/net/getnameinfo.c : 1.32 : 497 : 503 341.1Slukemdescr 351.1Slukem Because the current implementation of IP over IEEE1394, the 361.1Slukem fw device address contains more than just the IEEE1394 EUI-64. 371.1Slukem So when printing out IEEE1394 addresses, ignore the extra stuff. 381.1Slukemkcah 391.1Slukem 401.1Slukemhack xterm vs. libterm 411.1Slukemmdate 01 Aug 2000 421.1Slukemwho jdc 431.1Slukemfile xsrc/xc/programs/xterm/main.c : 1.2 : 3609 : 3614 441.1Slukempr 10383 451.1Slukemdescr 461.1Slukem In order to extend the termcap string over 1023 bytes, a ZZ entry was 471.1Slukem introduced to point to a memory location containing the full entry. 481.1Slukem Without this hack, xterm will export a termcap containing the ZZ 491.1Slukem entry, which will then be ignored by libterm. As xterm modifies the 501.1Slukem exported termcap, this would cause those modifications to be ignored. 511.1Slukemkcah 521.1Slukem 531.1Slukemhack wi-at-big-endian-bus 541.1Slukemcdate 15 Mar 2002 551.1Slukemwho martin 561.1Slukemfile dev/ic/wireg.h 571.1Slukemdescr Add an option to access the underlying bus in big endian byte order 581.1Slukem to work around deficiencies in bus_space_{read,write}_* macros. 591.1Slukem Those don't allow the implementation of a proper pcmcia bus space 601.1Slukem tag. 611.1Slukemkcah 621.1Slukem 631.3Sjdchack specific knowledge of colours in curses code 641.3Sjdccdate Sun Apr 6 11:05:24 BST 2003 651.3Sjdcwho jdc 661.3Sjdcfile lib/libcurses/color.c : r1.24 671.3Sjdcdescr 681.3Sjdc Swap red/blue and yellow/cyan colours for COLOR_OTHER. 691.3Sjdc Fix is to enhance libtermcap to understand terminfo-style % sequences. 701.3Sjdc See also: 711.5Ssalo http://mail-index.NetBSD.org/tech-userlevel/2003/04/06/0000.html 721.4Sscwkcah 731.4Sscw 741.7Shehack Compensation for differing types of LINUX_USRSTACK and USRSTACK 751.7Shecdate 21 Aug 2003 761.7Shewho he 771.7Shefile sys/miscfs/procfs/procfs_linux.c : 1.14 781.7Shedescr 791.7She Not all ports have LINUX_USRSTACK and/or USRSTACK defined as 801.7She literals/constants, but refer to variables of a type which is 811.7She not "unsigned long", causing compilation of procfs_linux.c to 821.7She fail with "makes integer from pointer without a cast". This 831.7She is observed on e.g. the sun3 port. Ideally the "types" for 841.7She symbols should be consistent across all ports. 851.11Slukemkcah 861.11Slukem 871.13Ssimonb 881.21Smycroftport vax 891.21Smycroft 901.21Smycroft hack gcc 2.95/vax cannot handle __builtin_ffs() 911.21Smycroft cdate Fri Aug 23 21:31:15 CEST 2002 921.21Smycroft who ragge 931.21Smycroft file sys/lib/libkern/libkern.h : 1.42 941.21Smycroft descr 951.21Smycroft __builtin_ffs() is ifdef'd away if __vax__. Matt Thomas has 961.21Smycroft added this feature to GCC 3 so it can be removed when switching. 971.21Smycroft kcah 981.21Smycroft 991.21Smycroft hack gcc-vax-libbz2 1001.21Smycroft mdate 27 Jun 2002 1011.21Smycroft who thorpej 1021.21Smycroft file lib/libbz2/Makefile 1031.21Smycroft descr 1041.21Smycroft libbz2 is mis-compiled with optimization with GCC 2.95.3 1051.21Smycroft on VAX. -O0 works around this problem. 1061.21Smycroft kcah 1071.21Smycroft 1081.21Smycroft hack vax Toolchain bug 1091.21Smycroft cdate 28 Mar 2003 1101.21Smycroft who he 1111.21Smycroft file sys/arch/vax/vax/intvec.S : 1.5 1121.21Smycroft descr 1131.21Smycroft Workaround for PR toolchain/20924. The assembler apparently 1141.21Smycroft tries to range-check byte offsets when it doesn't have 1151.21Smycroft sufficient information to make that decision. 1161.21Smycroft Workaround uses `brw' instead of `brb' instruction. 1171.21Smycroft kcah 1181.21Smycroft 1191.21Smycroft 1201.21Smycroftport sparc64 1211.21Smycroft 1221.21Smycroft hack binutil-2.11-sparc64-pltrela 1231.21Smycroft mdate 14 Aug 2001 1241.21Smycroft who eeh 1251.21Smycroft file libexec/ld.elf_so/arch/sparc64/mdreloc.c 1261.21Smycroft descr 1271.21Smycroft The first for PLT entries are reserved. There is some 1281.21Smycroft disagreement whether they should have associated relocation 1291.21Smycroft entries. Both the SPARC 32-bit and 64-bit ELF specifications 1301.21Smycroft say that they should have relocation entries, but the 32-bit 1311.21Smycroft SPARC binutils do not generate them, and now the 64-bit SPARC 1321.21Smycroft binutils have stopped generating them too. 1331.21Smycroft 1341.21Smycroft To provide binary compatibility, we will check the first entry, 1351.21Smycroft if it is reserved it should not be of the type JMP_SLOT. If it 1361.21Smycroft is JMP_SLOT, then the 4 reserved entries were not generated and 1371.21Smycroft our index is 4 entries too far, so we frob the rela pointer. 1381.21Smycroft kcah 1391.21Smycroft 1401.21Smycroft 1411.21Smycroftport arm 1421.21Smycroft 1431.21Smycroft hack gcc-unsigned-compare 1441.21Smycroft cdate 09 Mar 2002 1451.21Smycroft mdate 18 Mar 2002 1461.21Smycroft who bjh21 1471.21Smycroft file dist/bind/lib/nameser/ns_parse.c : 1.3 1481.21Smycroft file dist/dhcp/minires/ns_parse.c : 1.3 1491.21Smycroft file dist/dhcp/omapip/result.c : 1.2 1501.21Smycroft file dist/dhcp/server/failover.c : 1.3 1511.21Smycroft file gnu/dist/gawk/eval.c : 1.4 1521.21Smycroft file gnu/dist/toolchain/bfd/bfd.c : 1.2 1531.21Smycroft file gnu/dist/toolchain/bfd/format.c : 1.2 1541.21Smycroft file gnu/dist/toolchain/gdb/target.c : 1.2 1551.21Smycroft file sys/kern/vfs_subr.c : 1.172 1561.21Smycroft descr When checking that a potentially-unsigned enum is >= 0, assign 1571.21Smycroft it to an int first. This is necessary to avoid "comparison is 1581.21Smycroft always true" warnings with -fshort-enums. Casting to an int 1591.21Smycroft really should be enough, but turns out not to be. 1601.21Smycroft kcah 1611.21Smycroft 1621.21Smycroft hack gcc 3.3.x bug 1631.21Smycroft cdate Sun Oct 26 01:50:16 UTC 2003 1641.21Smycroft who lukem 1651.21Smycroft file usr.bin/newsyslog/newsyslog.c 1661.21Smycroft pr 22986 23002 23217 1671.21Smycroft descr 1681.21Smycroft newsyslog fails with 1691.21Smycroft newsyslog: config line 7: bad flags 1701.21Smycroft solved when compiled with -fno-gcse-lm 1711.21Smycroft kcah 1721.21Smycroft 1731.21Smycroft hack gcc 3.3.2/arm sys/kern/uipc_socket.c 1741.21Smycroft cdate Tue Oct 28 18:02:16 GMT 2003 1751.21Smycroft who skrll 1761.21Smycroft file sys/conf/arm/Makefile.arm : 1.19 1771.21Smycroft descr 1781.21Smycroft Performing a build.sh tools on a CATS with NFS mounted 1791.21Smycroft obj and src directories results in "panic: receive 1" 1801.21Smycroft The hack is to compile sys/kern/uipc_socket.c with 1811.21Smycroft -fno-strict-aliasing 1821.21Smycroft kcah 1831.21Smycroft 1841.21Smycroft 1851.13Ssimonbport pc532 1861.16Scl 1871.21Smycroft hack egcs-pc532-ip6_mroute 1881.21Smycroft cdate 09 Jul 2002 1891.21Smycroft who simonb 1901.21Smycroft file sys/arch/pc532/conf/Makefile.pc532 : 1.70 1911.21Smycroft file sys/arch/pc532/conf/files.pc532 : 1.47 1921.21Smycroft descr 1931.21Smycroft egcs 1.1.2 gets an "internal error--insn does not satisfy its 1941.21Smycroft constraints" error compiling ip6_mroute.c with -O2 or greater. 1951.21Smycroft -O1 works around this problem. 1961.21Smycroft kcah 1971.21Smycroft 1981.21Smycroft hack gcc-pc532-duffs_device 1991.21Smycroft cdate Mon Oct 27 07:23:05 UTC 2003 2001.21Smycroft who simonb 2011.21Smycroft file sys/arch/pc532/conf/Makefile.pc532 : 1.71 2021.21Smycroft file sys/arch/pc532/conf/files.pc532 : 1.52 2031.21Smycroft pr GCC PR optimization/5230 2041.21Smycroft descr 2051.21Smycroft gcc incorrectly gives an "unreachable code at beginning of 2061.21Smycroft switch statement" for a Duff's device construct in 2071.21Smycroft arch/pc532/dev/ncr.c. There is no way to disable just this 2081.21Smycroft warning, so -Wno-error is turned on for this file. 2091.21Smycroft kcah 2101.21Smycroft 2111.21Smycroft 2121.21Smycroftport sh3 2131.21Smycroft 2141.21Smycroft hack gcc-sh3-sed 2151.21Smycroft mdate 23 Apr 2002 2161.21Smycroft who thorpej 2171.21Smycroft file usr.bin/sed/Makefile : 1.9 : 9 : 13 2181.21Smycroft descr 2191.21Smycroft The in-tree GCC 2.95.3-based compiler ICEs when building 2201.21Smycroft with optimization for SuperH. Hack is to biuld with -O0. 2211.21Smycroft kcah 2221.21Smycroft 2231.21Smycroft 2241.21Smycroftport sh5 2251.21Smycroft 2261.21Smycroft hack SuperH SH5 Toolchain Bugs 2271.21Smycroft cdate 11 Jul 2002 2281.21Smycroft who scw 2291.21Smycroft file usr.sbin/ndbootd/ndbootd.c : 1.6 2301.21Smycroft file usr.sbin/traceroute/traceroute.c : 1.48 2311.21Smycroft descr 2321.21Smycroft The SuperH SH5 toolchain (2.97-sh5-010522) gets an internal 2331.21Smycroft compiler error when assigning a bit-wise inverted value 2341.21Smycroft under some circumstances. 2351.21Smycroft Work around it by splitting the statement into two. 2361.21Smycroft kcah 2371.21Smycroft 2381.21Smycroft hack SuperH SH5 Toolchain Bugs 2391.21Smycroft cdate 17 May 2003 2401.21Smycroft who scw 2411.21Smycroft file lib/libpthread/pthread_lock.c : 1.7 2421.21Smycroft descr 2431.21Smycroft The SuperH SH5 toolchain generates incorrect PIC code when 2441.21Smycroft faced with a symbol which is declared extern, but has local 2451.21Smycroft scope due to being defined within an asm() statement (without 2461.21Smycroft being declared .globl in that statement). Work around it by 2471.21Smycroft adding the .globl. 2481.21Smycroft kcah 2491.17Smrg 2501.17Smrg 2511.17Smrgport sun2 2521.20Sskrll 2531.21Smycroft hack gcc 3.3.2/mdsetimage 2541.21Smycroft cdate Tue Oct 28 18:43:05 EST 2003 2551.21Smycroft who fredette 2561.21Smycroft file gnu/usr.sbin/mdsetimage/Makefile : 1.15 2571.21Smycroft descr 2581.21Smycroft mdsetimage.c causes cc1 to barf for as-yet unknown reasons. 2591.21Smycroft this hack ues -O0 to avoid the problem. 2601.21Smycroft kcah 2611.21Smycroft 2621.21Smycroft hack gcc 3.3.2/installboot/ffs.c 2631.21Smycroft cdate Tue Oct 28 18:43:05 EST 2003 2641.21Smycroft who fredette 2651.21Smycroft file usr.sbin/installboot/Makefile : 1.26 2661.21Smycroft descr 2671.21Smycroft ffs.c causes cc1 to barf for as-yet unknown reasons. this 2681.21Smycroft hack ues -O0 to avoid the problem. 2691.22Smrg kcah 2701.22Smrg 2711.22Smrg 2721.22Smrgport sun2, m68k 2731.22Smrg 2741.22Smrg hack gcc 3.3.2/gcc/ra-colorize.c 2751.22Smrg cdate Wed Nov 5 11:03:13 EST 2003 2761.22Smrg who mrg 2771.22Smrg file gnu/usr.bin/gcc/backend/Makefile : 1.11 2781.22Smrg descr 2791.22Smrg ra-colorize.c causes a whole spew of unreferenced .L* local 2801.22Smrg symbols. using -O0 (or even -O2 reportedly) avoids the problem. 2811.21Smycroft kcah 282