Created attachment 452154 [details] strace output With sandbox-2.10-r1 gcl emerged fine. With 2.10-r2 cp init_pcl_gcl.lsp foo echo " (in-package \"USER\")(system:save-system \"saved_pcl_gcl\")" >>foo ar x libpcl_gcl.a $(ar t libpcl_gcl.a |grep ^gcl_) /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/unixport/raw_pcl_gcl /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/unixport/ -libdir /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/ < foo Unrecoverable error: Segmentation violation.. /bin/sh: line 1: 22515 Aborted /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/unixport/raw_pcl_gcl /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/unixport/ -libdir /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/ < foo make[1]: *** [makefile:130: saved_pcl_gcl] Error 134 rm init_pcl_gcl.lsp raw_pcl_gcl make[1]: Leaving directory '/var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/unixport' make: *** [makefile:88: unixport/saved_pcl_gcl] Error 2 When I go to /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/unixport/ and run raw_gcl with *exactly the same parameters* from the command line, it succeeds and produces saved_gcl. I've modified unixport/makefile to run raw_gcl under gdb and produce backtrace. This is the result: (gdb) Starting program: /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/unixport/raw_gcl /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/unixport/ -libdir /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/ < foo [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib64/libthread_db.so.1". Program received signal SIGSEGV, Segmentation fault. 0x0000000000367ee0 in calloc () (gdb) #0 0x0000000000367ee0 in calloc () #1 0x00007ffff737f687 in ?? () from /lib64/libdl.so.2 #2 0x00007ffff737f158 in dlsym () from /lib64/libdl.so.2 #3 0x00007ffff7bc2fc8 in ?? () from /usr/lib64/libsandbox.so #4 0x00007ffff7bc1e4e in ?? () from /usr/lib64/libsandbox.so #5 0x00007ffff7bc802b in ?? () from /usr/lib64/libsandbox.so #6 0x00007ffff7bc08a7 in ?? () from /usr/lib64/libsandbox.so #7 0x00007ffff7bc0c98 in ?? () from /usr/lib64/libsandbox.so #8 0x00007ffff7bc5b9c in open () from /usr/lib64/libsandbox.so #9 0x0000000000312a89 in get_phys_pages_no_malloc () #10 0x0000000000312c08 in update_real_maxpage () #11 0x0000000000365aea in gcl_init_alloc () #12 0x0000000000366ba2 in malloc () #13 0x0000000000367ed0 in calloc () #14 0x00007ffff737f687 in ?? () from /lib64/libdl.so.2 #15 0x00007ffff737f158 in dlsym () from /lib64/libdl.so.2 #16 0x00007ffff7bc2fc8 in ?? () from /usr/lib64/libsandbox.so #17 0x00007ffff7bc1e4e in ?? () from /usr/lib64/libsandbox.so #18 0x00007ffff7bc802b in ?? () from /usr/lib64/libsandbox.so #19 0x00007ffff7bc08a7 in ?? () from /usr/lib64/libsandbox.so #20 0x00007ffff7bbfa26 in ?? () from /usr/lib64/libsandbox.so #21 0x00007ffff7de82da in ?? () from /lib64/ld-linux-x86-64.so.2 #22 0x00007ffff7de83eb in ?? () from /lib64/ld-linux-x86-64.so.2 ---Type <return> to continue, or q <return> to quit---Quit I've also modified unixport/makefile to run raw_gcl under strace, the result is attached. This is the letter I received from Camm Maguire, the author of gcl: Greetings! In the failing environment, there is apparently some wrapper to open() which is interfering with a read of /proc/meminfo: 17104 brk(0x800d12000) = 0x800d12000 17104 brk(0x800d16000) = 0x800d16000 17104 brk(0x800d18000) = 0x800d18000 17104 brk(0xd1b000) = 0xd1b000 17104 open("/proc/meminfo", O_RDONLY) = 3 17104 read(3, "MemTotal: 8074492 kB\nMemF"..., 4096) = 1011 17104 close(3) = 0 17104 brk(0xcd9e3000) = 0xcd9e3000 17104 stat("/proc/17104/exe", {st_mode=S_IFREG|0755, st_size=4446024, ...}) = 0 vs 16803 brk(0x800d0a000) = 0x800d0a000 16803 brk(0x800d12000) = 0x800d12000 16803 brk(0x800d16000) = 0x800d16000 16803 brk(0x800d18000) = 0x800d18000 16803 brk(0xd1b000) = 0xd1b000 16803 stat("/proc/meminfo", {st_mode=S_IFREG|0444, st_size=0, ...}) = 0 16803 --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0x1f} --- 16803 fstat(1, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 4), ...}) = 0 16803 write(1, "\nUnrecoverable error: ", 22) = 22 16803 write(1, "S", 1) = 1 16803 write(1, "egmentation violation.", 22) = 22 This method of determining the runtime available memory has been replaced in 2.6.13pre with (hopefully) more robust alternatives. But hopefully this should point out what is going on. FEATURES="-sandbox -usersandbox" emerge gcl solves the problem.
pretty sure i already fixed this with sandbox-2.11 via https://crbug.com/586444
(In reply to SpanKY from comment #1) > pretty sure i already fixed this with sandbox-2.11 via > https://crbug.com/586444 It's masked: # Patrick Lauer (30 Mar 2015) # Segfaults when creating directories #578582 =sys-apps/sandbox-2.11*
(In reply to Andrey Grozin from comment #2) which Patrick never told anyone in his standard style. i'll fix that.
After upgrade to sandbox-2.11-r2 nothing changed. The output of gdb is (gdb) Starting program: /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/unixport/raw_gcl /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/unixport/ -libdir /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/ < foo [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib64/libthread_db.so.1". Program received signal SIGSEGV, Segmentation fault. 0x0000000000367ee0 in calloc () (gdb) #0 0x0000000000367ee0 in calloc () #1 0x00007ffff737f687 in ?? () from /lib64/libdl.so.2 #2 0x00007ffff737f158 in dlsym () from /lib64/libdl.so.2 #3 0x00007ffff7bc2f08 in ?? () from /usr/lib64/libsandbox.so #4 0x00007ffff7bc201e in ?? () from /usr/lib64/libsandbox.so #5 0x00007ffff7bc898b in ?? () from /usr/lib64/libsandbox.so #6 0x00007ffff7bc08e7 in ?? () from /usr/lib64/libsandbox.so #7 0x00007ffff7bc0cd8 in ?? () from /usr/lib64/libsandbox.so #8 0x00007ffff7bc650c in open () from /usr/lib64/libsandbox.so #9 0x0000000000312a89 in get_phys_pages_no_malloc () #10 0x0000000000312c08 in update_real_maxpage () #11 0x0000000000365aea in gcl_init_alloc () #12 0x0000000000366ba2 in malloc () #13 0x0000000000367ed0 in calloc () #14 0x00007ffff737f687 in ?? () from /lib64/libdl.so.2 #15 0x00007ffff737f158 in dlsym () from /lib64/libdl.so.2 #16 0x00007ffff7bc2f08 in ?? () from /usr/lib64/libsandbox.so #17 0x00007ffff7bc201e in ?? () from /usr/lib64/libsandbox.so #18 0x00007ffff7bc898b in ?? () from /usr/lib64/libsandbox.so #19 0x00007ffff7bc08e7 in ?? () from /usr/lib64/libsandbox.so #20 0x00007ffff7bbfa66 in ?? () from /usr/lib64/libsandbox.so #21 0x00007ffff7de82ba in ?? () from /lib64/ld-linux-x86-64.so.2 #22 0x00007ffff7de83cb in ?? () from /lib64/ld-linux-x86-64.so.2 ---Type <return> to continue, or q <return> to quit---Quit (gdb) quit I can get the strace output if needed.
This is still an issue with dev-lisp/gcl-2.6.12 and sys-apps/sandbox-2.18 ar: creating libpre_gcl.a cp ../o/gcllib.a libgclp.a ranlib libgclp.a cp init_pre_gcl.lsp.in init_pre_gcl.lsp.tmp cat init_pre_gcl.lsp.tmp | sed \ -e "s#@LI-VERS@#(`cat ../majvers`.`cat ../minvers`) `date`#1" \ -e "s#@LI-EXTVERS@#`cat ../minvers | cut -f2 -d.`#1" \ -e "s#@LI-MINVERS@#`cat ../minvers | cut -f1 -d.`#1" \ -e "s#@LI-MAJVERS@#`cat ../majvers`#1" \ -e "s#@LI-CC@#\"gcc -c -Os -pipe -march=native -mtune=native -Wdate-time -Wno-error=implicit-function-declaration -Wformat -Wformat-security -Wl,-O1 -Wl,--as-needed -Wl,-O2 -Wl,-fuse-ld=bfd -Wl,--hash-style=gnu -Wl,--gc-sections -fgnu89-inline -D_REENTRANT -DREENTRANT -Wall -fsigned-char -Wno-unused-but-set-variable -pipe \"#1" \ -e "s#@LI-LD@#\"gcc -Wl,-O1 -Wl,--as-needed -Wl,-O2 -Wl,--as-needed -Wl,-fuse-ld=bfd -Wl,--hash-style=gnu -Wl,--gc-sections -Wl,-T ../unixport/gcl.script -o \"#1" \ -e "s#@LI-LD-LIBS@#\" -lpre_gcl -lm -ldl -lgmp -lreadline -lc -lgclp\"#1" \ -e "s#@LI-OPT-THREE@#\"-O3 \"#1" \ -e "s#@LI-OPT-TWO@#\"-O\"#1" \ -e "s#@LI-INIT-LSP@#\"init_pre_gcl.lsp\"#1" >init_pre_gcl.lsp touch raw_pre_gcl_map gcc -Wl,-O1 -Wl,--as-needed -Wl,-O2 -Wl,--as-needed -Wl,-fuse-ld=bfd -Wl,--hash-style=gnu -Wl,--gc-sections -Wl,-T ../unixport/gcl.script -o raw_pre_gcl -L. -Wl,-Map raw_pre_gcl_map -lpre_gcl -lm -ldl -lgmp -lreadline -lc -lgclp chcon -t gcl_exec_t raw_pre_gcl || true chcon: failed to get security context of 'raw_pre_gcl': Operation not supported PATH=/usr/bin:$PATH gcc msys.c -o msys # Unix binary if running wine cp init_pre_gcl.lsp foo echo " (in-package \"USER\")(system:save-system \"saved_pre_gcl\")" >>foo ar x libpre_gcl.a $(ar t libpre_gcl.a |grep ^gcl_) /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/unixport/raw_pre_gcl /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/unixport/ -libdir /var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/ < foo Unrecoverable error: Segmentation violation.. make[1]: *** [makefile:130: saved_pre_gcl] Aborted (core dumped) rm raw_pre_gcl make[1]: Leaving directory '/var/tmp/portage/dev-lisp/gcl-2.6.12/work/gcl/unixport' make: *** [makefile:78: unixport/saved_pre_gcl] Error 2 rm h/mcompdefs.h * ERROR: dev-lisp/gcl-2.6.12::gentoo failed (compile phase): * emake failed
I'm afraid that unless someone can figure it out and/or supply a patch, we should simply lastrite gcl. It's not the only lisp compiler out there, and FWICS no packages depend on it exclusively.
AFAIK this isn't an issue anymore.
