Additional info on modpost segfault
in [Kernel]
|
Prev: modpost segfaults with 32bit compile on 64 bit version of Fedora 13
Next: [PATCH] x86: Look for IA32_ENERGY_PERF_BIAS support - v3
From: Alan on 8 Jun 2010 14:30 On Tue, 2010-06-08 at 13:51 +0800, Américo Wang wrote: > On Mon, Jun 07, 2010 at 09:59:39AM -0700, Alan wrote: > >On Fri, 2010-06-04 at 09:22 +0200, Michal Marek wrote: > >> On 4.6.2010 06:51, Américo Wang wrote: > >> > On Thu, Jun 03, 2010 at 08:10:30PM -0700, alan wrote: > >> >> Missed adding the actual segfault message: > >> >> > >> >> LD drivers/usb/built-in.o > >> >> LD drivers/built-in.o > >> >> LD vmlinux.o > >> >> MODPOST vmlinux.o > >> >> /bin/sh: line 1: 20665 Segmentation fault (core dumped) > >> >> scripts/mod/modpost -o > >> >> /home/alan/GitTrees/linux-2.6-mid-ref/Module.symvers -S vmlinux.o > >> >> make[1]: *** [vmlinux.o] Error 139 > >> >> make: *** [vmlinux.o] Error 2 > >> >> > >> >> I have looked at the gcc 4.4.4 changelog and I can't see anything > >> >> that should cause this. > >> >> > >> > > >> > Hmm, you need to find which program segfaults here. > >> > >> It's the modpost command run on vmlinux.o. Alan, can you try > >> $ gdb --args scripts/mod/modpost -o Module.symvers -S vmlinux.o > >> (gdb) r > >> (wait for the segfault) > >> (gdb) bt full > >> > >> and post the backtrace? > > > >Don't know if this will help much. > > > >This GDB was configured as "x86_64-redhat-linux-gnu". > >For bug reporting instructions, please see: > ><http://www.gnu.org/software/gdb/bugs/>... > >Reading symbols > >from /home/alan/GitTrees/linux-2.6-mid-ref/scripts/mod/modpost...(no > >debugging symbols found)...done. > >(gdb) r > >Starting > >program: /home/alan/GitTrees/linux-2.6-mid-ref/scripts/mod/modpost -o > >Module.symvers -S vmlinux.o > > > >Program received signal SIGSEGV, Segmentation fault. > >0x0000000000403711 in main () > >(gdb) bt full > >#0 0x0000000000403711 in main () > >No symbol table info available. > > > >Trying to get it to compile with debugging info. > > > > Try to append "-g" to HOSTCFLAGS in the top Makefile. ;) Thanks. Tried CFLAGS and that did not work... Here is the backtrace: (gdb) bt full #0 read_symbols (argc=5, argv=0x7fffffffe198) at scripts/mod/modpost.c:1564 license = <value optimized out> info = {size = 209695023, hdr = 0x7fffeb263000, sechdrs = 0x7ffff48fcc2c, symtab_start = 0x7ffff7863e2c, symtab_stop = 0x7ffff794341c, export_sec = 37, export_unused_sec = 0, export_gpl_sec = 48, export_unused_gpl_sec = 0, export_gpl_future_sec = 0, strtab = 0x7ffff794341c "", modinfo = 0x0, modinfo_len = 0} sym = <value optimized out> symname = <value optimized out> version = <value optimized out> mod = 0x610010 #1 main (argc=5, argv=0x7fffffffe198) at scripts/mod/modpost.c:1999 mod = <value optimized out> buf = {p = 0x0, pos = 0, size = 0} kernel_read = <value optimized out> module_read = <value optimized out> dump_write = 0x7fffffffe4ed "Module.symvers" opt = <value optimized out> err = <value optimized out> extsym_iter = <value optimized out> extsym_start = <value optimized out> The line is does not like is: read_symbols(argv[optind++]); Not certain why... -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Américo Wang on 9 Jun 2010 04:40 On Tue, Jun 08, 2010 at 11:24:18AM -0700, Alan wrote: >On Tue, 2010-06-08 at 13:51 +0800, Américo Wang wrote: >> On Mon, Jun 07, 2010 at 09:59:39AM -0700, Alan wrote: >> >On Fri, 2010-06-04 at 09:22 +0200, Michal Marek wrote: >> >> On 4.6.2010 06:51, Américo Wang wrote: >> >> > On Thu, Jun 03, 2010 at 08:10:30PM -0700, alan wrote: >> >> >> Missed adding the actual segfault message: >> >> >> >> >> >> LD drivers/usb/built-in.o >> >> >> LD drivers/built-in.o >> >> >> LD vmlinux.o >> >> >> MODPOST vmlinux.o >> >> >> /bin/sh: line 1: 20665 Segmentation fault (core dumped) >> >> >> scripts/mod/modpost -o >> >> >> /home/alan/GitTrees/linux-2.6-mid-ref/Module.symvers -S vmlinux.o >> >> >> make[1]: *** [vmlinux.o] Error 139 >> >> >> make: *** [vmlinux.o] Error 2 >> >> >> >> >> >> I have looked at the gcc 4.4.4 changelog and I can't see anything >> >> >> that should cause this. >> >> >> >> >> > >> >> > Hmm, you need to find which program segfaults here. >> >> >> >> It's the modpost command run on vmlinux.o. Alan, can you try >> >> $ gdb --args scripts/mod/modpost -o Module.symvers -S vmlinux.o >> >> (gdb) r >> >> (wait for the segfault) >> >> (gdb) bt full >> >> >> >> and post the backtrace? >> > >> >Don't know if this will help much. >> > >> >This GDB was configured as "x86_64-redhat-linux-gnu". >> >For bug reporting instructions, please see: >> ><http://www.gnu.org/software/gdb/bugs/>... >> >Reading symbols >> >from /home/alan/GitTrees/linux-2.6-mid-ref/scripts/mod/modpost...(no >> >debugging symbols found)...done. >> >(gdb) r >> >Starting >> >program: /home/alan/GitTrees/linux-2.6-mid-ref/scripts/mod/modpost -o >> >Module.symvers -S vmlinux.o >> > >> >Program received signal SIGSEGV, Segmentation fault. >> >0x0000000000403711 in main () >> >(gdb) bt full >> >#0 0x0000000000403711 in main () >> >No symbol table info available. >> > >> >Trying to get it to compile with debugging info. >> > >> >> Try to append "-g" to HOSTCFLAGS in the top Makefile. ;) > >Thanks. Tried CFLAGS and that did not work... > >Here is the backtrace: >(gdb) bt full >#0 read_symbols (argc=5, argv=0x7fffffffe198) at >scripts/mod/modpost.c:1564 > license = <value optimized out> > info = {size = 209695023, hdr = 0x7fffeb263000, > sechdrs = 0x7ffff48fcc2c, symtab_start = 0x7ffff7863e2c, > symtab_stop = 0x7ffff794341c, export_sec = 37, > export_unused_sec = 0, export_gpl_sec = 48, > export_unused_gpl_sec = 0, export_gpl_future_sec = 0, > strtab = 0x7ffff794341c "", modinfo = 0x0, modinfo_len = 0} > sym = <value optimized out> > symname = <value optimized out> > version = <value optimized out> > mod = 0x610010 >#1 main (argc=5, argv=0x7fffffffe198) at scripts/mod/modpost.c:1999 > mod = <value optimized out> > buf = {p = 0x0, pos = 0, size = 0} > kernel_read = <value optimized out> > module_read = <value optimized out> > dump_write = 0x7fffffffe4ed "Module.symvers" > opt = <value optimized out> > err = <value optimized out> > extsym_iter = <value optimized out> > extsym_start = <value optimized out> > >The line is does not like is: > > read_symbols(argv[optind++]); > >Not certain why... Hmm, it seems the segfault happens at 'license = get_modinfo(...)'? I can't spot any bug around that. If I were you, I would do a step by step debug with gdb. Thanks! -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Krzysztof Halasa on 10 Jun 2010 19:10 Alan <alan(a)clueserver.org> writes: > program: /home/alan/GitTrees/linux-2.6-mid-ref/scripts/mod/modpost -o > Module.symvers -S vmlinux.o > > Program received signal SIGSEGV, Segmentation fault. It just hit me. It's the offset calculation in reloc_location() which overflows: return (void *)elf->hdr + sechdrs[section].sh_offset + (r->r_offset - sechdrs[section].sh_addr); E.g. for the first rodata r entry: r->r_offset < sechdrs[section].sh_addr and the expression in the parenthesis produces 0xFFFFFFE0 or something equally wise. Does the attached patch fix it? Signed-off-by: Krzysztof Hałasa <khc(a)pm.waw.pl> --- a/scripts/mod/modpost.c +++ b/scripts/mod/modpost.c @@ -1317,8 +1317,8 @@ static unsigned int *reloc_location(struct elf_info *elf, Elf_Shdr *sechdrs = elf->sechdrs; int section = sechdr->sh_info; return (void *)elf->hdr + sechdrs[section].sh_offset + - (r->r_offset - sechdrs[section].sh_addr); + r->r_offset - sechdrs[section].sh_addr; } static int addend_386_rel(struct elf_info *elf, Elf_Shdr *sechdr, Elf_Rela *r) -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: alan on 10 Jun 2010 20:10 On Fri, 11 Jun 2010, Krzysztof Halasa wrote: > Alan <alan(a)clueserver.org> writes: > >> program: /home/alan/GitTrees/linux-2.6-mid-ref/scripts/mod/modpost -o >> Module.symvers -S vmlinux.o >> >> Program received signal SIGSEGV, Segmentation fault. > > It just hit me. > It's the offset calculation in reloc_location() which overflows: > return (void *)elf->hdr + sechdrs[section].sh_offset + > (r->r_offset - sechdrs[section].sh_addr); > > E.g. for the first rodata r entry: > r->r_offset < sechdrs[section].sh_addr > and the expression in the parenthesis produces 0xFFFFFFE0 or something > equally wise. > > Does the attached patch fix it? YES! Thank you! Now the big question is why does this compile on older versions of gcc? This needs to get added into 2.6.35-rc2. > > Signed-off-by: Krzysztof Ha��asa <khc(a)pm.waw.pl> > > --- a/scripts/mod/modpost.c > +++ b/scripts/mod/modpost.c > @@ -1317,8 +1317,8 @@ static unsigned int *reloc_location(struct elf_info *elf, > Elf_Shdr *sechdrs = elf->sechdrs; > int section = sechdr->sh_info; > > return (void *)elf->hdr + sechdrs[section].sh_offset + > - (r->r_offset - sechdrs[section].sh_addr); > + r->r_offset - sechdrs[section].sh_addr; > } > > static int addend_386_rel(struct elf_info *elf, Elf_Shdr *sechdr, Elf_Rela *r) > -- > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo(a)vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > Please read the FAQ at http://www.tux.org/lkml/ > -- Truth is stranger than fiction because fiction has to make sense.
From: H. Peter Anvin on 11 Jun 2010 16:50
Michal, are you sending this to Linus? -hpa On 06/10/2010 04:08 PM, Krzysztof Halasa wrote: > Alan <alan(a)clueserver.org> writes: > >> program: /home/alan/GitTrees/linux-2.6-mid-ref/scripts/mod/modpost -o >> Module.symvers -S vmlinux.o >> >> Program received signal SIGSEGV, Segmentation fault. > > It just hit me. > It's the offset calculation in reloc_location() which overflows: > return (void *)elf->hdr + sechdrs[section].sh_offset + > (r->r_offset - sechdrs[section].sh_addr); > > E.g. for the first rodata r entry: > r->r_offset < sechdrs[section].sh_addr > and the expression in the parenthesis produces 0xFFFFFFE0 or something > equally wise. > > Does the attached patch fix it? > > Signed-off-by: Krzysztof Hałasa <khc(a)pm.waw.pl> > > --- a/scripts/mod/modpost.c > +++ b/scripts/mod/modpost.c > @@ -1317,8 +1317,8 @@ static unsigned int *reloc_location(struct elf_info *elf, > Elf_Shdr *sechdrs = elf->sechdrs; > int section = sechdr->sh_info; > > return (void *)elf->hdr + sechdrs[section].sh_offset + > - (r->r_offset - sechdrs[section].sh_addr); > + r->r_offset - sechdrs[section].sh_addr; > } > > static int addend_386_rel(struct elf_info *elf, Elf_Shdr *sechdr, Elf_Rela *r) > -- > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo(a)vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > Please read the FAQ at http://www.tux.org/lkml/ -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/ |