From: Andi Kleen <ak@xxxxxxxxxxxxxxx> kallsyms currenly links the kernel upto three times (in addition to another one for modpost checks) Linking can be a quite slow operation, especially when the kernel has a lot of debug information (lots of IO), or Link Time Optimization is used. Final linking is also a non parallelizable bottlneck, so it's always good to do it less. Use a different kallsyms method to avoid this: - generate a initial kallsyms table from the top level object files - This table is usually a super set of the final table, but without final addresses and some extra symbols (e.g. discard and local symbols) - Use this table to link the vmlinux - Then generate a new kallsyms table with padding so that all symbols stay at the same offsets. This works because the new table is smaller or the same than the original one. We let the first kallsyms generate a padding file and then use it on the next link. - Then finally patch in the new table into the vmlinux The size difference between the two tables is typically small, so the additional padding is not a problem (a few hundred bytes in my kernels) Right now we still do two links. One to generate the kernel, and another one to generate the vmlinux.o for modpost. On my slowish laptop this cuts down the final serialized phase of a moderate size kernel build (just relinking vmlinux) by 1/3, from ~30s to 20s Tested on x86, tests on other architectures would be appreciated. Signed-off-by: Andi Kleen <ak@xxxxxxxxxxxxxxx> --- scripts/Makefile | 2 +- scripts/elf_file_offset | 24 +++++++++++++ scripts/kallsyms.c | 10 ++++-- scripts/link-vmlinux.sh | 92 ++++++++++++++++++++----------------------------- scripts/patchfile.c | 81 +++++++++++++++++++++++++++++++++++++++++++ 5 files changed, 151 insertions(+), 58 deletions(-) create mode 100644 scripts/elf_file_offset create mode 100644 scripts/patchfile.c diff --git a/scripts/Makefile b/scripts/Makefile index 01e7adb..9c84464 100644 --- a/scripts/Makefile +++ b/scripts/Makefile @@ -10,7 +10,7 @@ HOST_EXTRACFLAGS += -I$(srctree)/tools/include -hostprogs-$(CONFIG_KALLSYMS) += kallsyms +hostprogs-$(CONFIG_KALLSYMS) += kallsyms patchfile hostprogs-$(CONFIG_LOGO) += pnmtologo hostprogs-$(CONFIG_VT) += conmakehash hostprogs-$(CONFIG_IKCONFIG) += bin2c diff --git a/scripts/elf_file_offset b/scripts/elf_file_offset new file mode 100644 index 0000000..2c8ee49 --- /dev/null +++ b/scripts/elf_file_offset @@ -0,0 +1,24 @@ +#!/bin/bash +# find the file offset of a section in a ELF file +# objdump --section-headers elf-file | +# gawk -f elf_file_offset filesize=SIZE section=SECTIONNAME +# gawk needed for strtonum() +#Idx Name Size VMA LMA File off Algn +# 4 .kallsyms 001fd648 ffffffff81b1c068 0000000001b1c068 00d1c068 2**3 + +$2 == section { + old = strtonum("0x" $3) + new = strtonum(filesize) + if (old < new) { + print "Not enough padding in vmlinux for new kallsyms, missing",new-old > "/dev/stderr" + print "Please lower (=increase) PAD_RATIO in kallsyms.c" + exit 1 + } + print "0x" $6 + # XXX doesn't exit in gawk 4.1.0 ?!? + #exit(0) +} +#END { +# print section " not found" > "/dev/stderr" +# exit 1 +#} diff --git a/scripts/kallsyms.c b/scripts/kallsyms.c index 293c9eb..566e53e 100644 --- a/scripts/kallsyms.c +++ b/scripts/kallsyms.c @@ -192,15 +192,19 @@ static int symbol_valid_tr(struct sym_entry *s) { size_t i; struct text_range *tr; + int valid = 0; for (i = 0; i < ARRAY_SIZE(text_ranges); ++i) { tr = &text_ranges[i]; + if (tr->start && tr->end) + valid++; + if (s->addr >= tr->start && s->addr <= tr->end) return 1; } - return 0; + return valid ? 0 : 1; } static int symbol_valid(struct sym_entry *s) @@ -242,11 +246,13 @@ static int symbol_valid(struct sym_entry *s) * the kallsyms data are added. If these symbols move then * they may get dropped in pass 2, which breaks the kallsyms * rules. + * But don't do this for predicted fake symbols with 0 value. */ - if ((s->addr == text_range_text->end && + if (((s->addr == text_range_text->end && strcmp((char *)s->sym + offset, text_range_text->etext)) || (s->addr == text_range_inittext->end && strcmp((char *)s->sym + offset, text_range_inittext->etext))) + && text_range_text->end != 0) return 0; } diff --git a/scripts/link-vmlinux.sh b/scripts/link-vmlinux.sh index 2dcb377..b299fdd 100644 --- a/scripts/link-vmlinux.sh +++ b/scripts/link-vmlinux.sh @@ -1,4 +1,4 @@ -#!/bin/sh +#!/bin/bash # # link vmlinux # @@ -85,11 +85,13 @@ kallsyms() if [ -n "${CONFIG_ARM}" ] && [ -n "${CONFIG_PAGE_OFFSET}" ]; then kallsymopt="${kallsymopt} --page-offset=$CONFIG_PAGE_OFFSET" fi + kallsymopt="${kallsymopt} $3" local aflags="${KBUILD_AFLAGS} ${KBUILD_AFLAGS_KERNEL} \ ${NOSTDINC_FLAGS} ${LINUXINCLUDE} ${KBUILD_CPPFLAGS}" ${NM} -n ${1} | \ + awk 'NF == 3 { print}' | scripts/kallsyms ${kallsymopt} | \ ${CC} ${aflags} -c -o ${2} -x assembler-with-cpp - } @@ -166,51 +168,43 @@ ${MAKE} -f "${srctree}/scripts/Makefile.build" obj=init kallsymso="" kallsyms_vmlinux="" -if [ -n "${CONFIG_KALLSYMS}" ]; then - - # kallsyms support - # Generate section listing all symbols and add it into vmlinux - # It's a three step process: - # 1) Link .tmp_vmlinux1 so it has all symbols and sections, - # but __kallsyms is empty. - # Running kallsyms on that gives us .tmp_kallsyms1.o with - # the right size - # 2) Link .tmp_vmlinux2 so it now has a __kallsyms section of - # the right size, but due to the added section, some - # addresses have shifted. - # From here, we generate a correct .tmp_kallsyms2.o - # 2a) We may use an extra pass as this has been necessary to - # woraround some alignment related bugs. - # KALLSYMS_EXTRA_PASS=1 is used to trigger this. - # 3) The correct ${kallsymso} is linked into the final vmlinux. - # - # a) Verify that the System.map from vmlinux matches the map from - # ${kallsymso}. - - kallsymso=.tmp_kallsyms2.o - kallsyms_vmlinux=.tmp_vmlinux2 - - # step 1 - vmlinux_link "" .tmp_vmlinux1 - kallsyms .tmp_vmlinux1 .tmp_kallsyms1.o - - # step 2 - vmlinux_link .tmp_kallsyms1.o .tmp_vmlinux2 - kallsyms .tmp_vmlinux2 .tmp_kallsyms2.o - - # step 2a - if [ -n "${KALLSYMS_EXTRA_PASS}" ]; then - kallsymso=.tmp_kallsyms3.o - kallsyms_vmlinux=.tmp_vmlinux3 - - vmlinux_link .tmp_kallsyms2.o .tmp_vmlinux3 - - kallsyms .tmp_vmlinux3 .tmp_kallsyms3.o - fi + +if [ -n "${CONFIG_KALLSYMS}" ] ; then + # Generate kallsyms from the top level object files + # this is slightly off, and has wrong addresses, + # but gives us the conservative max length of the kallsyms + # table to link in something with the size. + info KALLSYMS1 .tmp_kallsyms1.o + kallsyms "${KBUILD_VMLINUX_INIT} ${KBUILD_VMLINUX_MAIN}" \ + .tmp_kallsyms1.o \ + "--pad-file=.kallsyms_pad" + kallsymsso=.tmp_kallsyms1.o fi info LD vmlinux -vmlinux_link "${kallsymso}" vmlinux +vmlinux_link "${kallsymsso}" vmlinux +if [ -n "${CONFIG_KALLSYMS}" ] ; then + # Now regenerate the kallsyms table and patch it into the + # previously linked file. We tell kallsyms to pad it + # to the previous length, so that no symbol changes. + info KALLSYMS2 .tmp_kallsyms2.o + kallsyms vmlinux .tmp_kallsyms2.o $(<.kallsyms_pad) + + info OBJCOPY .tmp_kallsyms2.bin + ${OBJCOPY} -O binary .tmp_kallsyms2.o .tmp_kallsyms2.bin + + info PATCHFILE vmlinux + EF=scripts/elf_file_offset + if [ ! -r $EF ] ; then EF=source/$EF ; fi + OFF=$(${OBJDUMP} --section-headers vmlinux | + gawk -f $EF \ + -v section=.kallsyms -v filesize=$(stat -c%s .tmp_kallsyms2.bin) ) + if [ -z "$OFF" ] ; then + echo "Cannot find .kallsyms section in vmlinux binary" + exit 1 + fi + scripts/patchfile vmlinux $OFF .tmp_kallsyms2.bin +fi if [ -n "${CONFIG_BUILDTIME_EXTABLE_SORT}" ]; then info SORTEX vmlinux @@ -220,17 +214,5 @@ fi info SYSMAP System.map mksysmap vmlinux System.map -# step a (see comment above) -if [ -n "${CONFIG_KALLSYMS}" ]; then - mksysmap ${kallsyms_vmlinux} .tmp_System.map - - if ! cmp -s System.map .tmp_System.map; then - echo >&2 Inconsistent kallsyms data - echo >&2 Try "make KALLSYMS_EXTRA_PASS=1" as a workaround - cleanup - exit 1 - fi -fi - # We made a new kernel - delete old version file rm -f .old_version diff --git a/scripts/patchfile.c b/scripts/patchfile.c new file mode 100644 index 0000000..1a3414d --- /dev/null +++ b/scripts/patchfile.c @@ -0,0 +1,81 @@ +/* Patch file at specific offset + * patchfile file-to-patch offset patch-file [len-of-patch] + */ +#define _GNU_SOURCE 1 +#include <sys/mman.h> +#include <unistd.h> +#include <sys/fcntl.h> +#include <sys/stat.h> +#include <stdio.h> +#include <unistd.h> +#include <stdlib.h> + +#define ROUNDUP(x, y) (((x) + (y) - 1) & ~((y) - 1)) + +static void *mmapfile(char *file, size_t *size) +{ + int pagesize = sysconf(_SC_PAGESIZE); + int fd = open(file, O_RDONLY); + void *res = NULL; + struct stat st; + + *size = 0; + if (fd < 0) + return NULL; + if (fstat(fd, &st) >= 0) { + *size = st.st_size; + res = mmap(NULL, ROUNDUP(st.st_size, pagesize), + PROT_READ, MAP_SHARED, + fd, 0); + if (res == (void *)-1) + res = NULL; + } + close(fd); + return res; +} + +static void usage(void) +{ + fprintf(stderr, "Usage: patchfile file-to-patch offset file-to-patch-in\n"); + exit(1); +} + +static size_t get_num(char *s) +{ + char *endp; + size_t v = strtoul(s, &endp, 0); + if (s == endp) + usage(); + return v; +} + +int main(int ac, char **av) +{ + char *patch; + size_t patchsize; + int infd; + size_t offset; + + if (ac != 5 && ac != 4) + usage(); + offset = get_num(av[2]); + patch = mmapfile(av[3], &patchsize); + if (av[4]) { + size_t newsize = get_num(av[4]); + if (newsize > patchsize) + fprintf(stderr, "kallsyms: warning, size larger than patch\n"); + if (newsize < patchsize) + patchsize = newsize; + } + infd = open(av[1], O_RDWR); + if (infd < 0) { + fprintf(stderr, "Cannot open %s\n", av[1]); + exit(1); + } + if (pwrite(infd, patch, patchsize, offset) != patchsize) { + fprintf(stderr, "Cannot write patch to %s\n", av[1]); + exit(1); + } + close(infd); + return 0; +} -- 1.8.5.2 -- To unsubscribe from this list: send the line "unsubscribe linux-kbuild" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html