On Thu, 30 Dec 2010 06:16:17 +0900 HATAYAMA Daisuke <d.hatayama@xxxxxxxxx> wrote: > This patchset changes a position of section header table, if exists, > from the last to the next to ELF header. According to ELF > specification, the order of ELF component layout is not specified > except for ELF header. See: > > http://www.sco.com/developers/gabi/latest/ch4.intro.html#file_format > > The merits are: > - reducing tracing memory maps from 3 times to 2 times. > - simple offset handling that makes the code easier to read. > Hmm ? What was problem ? And new layout can be handled by usual tools ? As gdb, readelf, objdump...etc. > arch/ia64/kernel/elfcore.c | 16 ----- > arch/um/sys-i386/elfcore.c | 16 ----- > fs/binfmt_elf.c | 136 +++++++++++++++++------------------------- > fs/binfmt_elf_fdpic.c | 141 ++++++++++++++++++-------------------------- > include/linux/elfcore.h | 1 - > kernel/elfcore.c | 5 -- > 6 files changed, 111 insertions(+), 204 deletions(-) > > I built and tested this patchset on x86_64. I also built it on ia64, > um-i386 and frv using cross compilers to cover the range. > > Here's a program useful to generate ELF core with many program header > entries, and three steps to produce such ELF core: > > $ sysctl vm.max_map_count=70000 > $ ulimit -c unlimted > $ mkmmap 65535 > What difference will be seen between before/after your patch ? coredump will be quicker ? Thanks, -Kame -- To unsubscribe from this list: send the line "unsubscribe linux-arch" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
