| Message ID | 20140712202449.16082.46103.stgit@zurg (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi Konstantin. On Sun, Jul 13, 2014 at 12:24:50AM +0400, Konstantin Khlebnikov wrote: > This patch adds option for collecting stack usage statistics > via gcc option '-fstack-usage' (needs gcc 4.6 or newer). > For each .o file gcc dumps stack-frame sizes into .su file. > > File format: > <file>:<line>:<column>:<function> <size> <qualifiers> > > select.c:870:5:do_sys_poll 1040 static > rtnetlink.c:1880:12:rtnl_newlink 552 dynamic > random.c:1334:1:random_read 664 dynamic,bounded > > Quote from the gcc manpage about qualifiers: > > The qualifier "static" means that the function > manipulates the stack statically: a fixed number of bytes > are allocated for the frame on function entry and > released on function exit; no stack adjustments are > otherwise made in the function. The second field is this > fixed number of bytes. > > The qualifier "dynamic" means that the function > manipulates the stack dynamically: in addition to the > static allocation described above, stack adjustments are > made in the body of the function, for example to push/pop > arguments around function calls. If the qualifier > "bounded" is also present, the amount of these > adjustments is bounded at compile time and the second > field is an upper bound of the total amount of stack used > by the function. If it is not present, the amount of > these adjustments is not bounded at compile time and the > second field only represents the bounded part. > > In comparison to the scripts/checkstack.pl this method is more accurate. > It takes into account function arguments, push/pop pairs, frame > pointers and saved return address. It gives information about all > functions except purely assember (for example sha1_transform_avx2). > > Also this patch adds make target 'stack-usage' which combines all .su > files into one sorted 'stack-usage'. Also prints top 100 and all unbounded > dynamic stack frames. > > Signed-off-by: Konstantin Khlebnikov <koct9i@gmail.com> > --- > .gitignore | 3 +++ > Makefile | 23 ++++++++++++++++++++++- > lib/Kconfig.debug | 9 +++++++++ > 3 files changed, 34 insertions(+), 1 deletion(-) > > diff --git a/.gitignore b/.gitignore > index f4c0b09..96b8984 100644 > --- a/.gitignore > +++ b/.gitignore > @@ -85,6 +85,9 @@ GTAGS > *~ > \#*# > > +*.su > +/stack-usage > + > # > # Leavings from module signing > # > diff --git a/Makefile b/Makefile > index 2167084..29fe588 100644 > --- a/Makefile > +++ b/Makefile > @@ -635,6 +635,11 @@ ifneq ($(CONFIG_FRAME_WARN),0) > KBUILD_CFLAGS += $(call cc-option,-Wframe-larger-than=${CONFIG_FRAME_WARN}) > endif > > +ifdef CONFIG_CC_STACK_USAGE > +KBUILD_CFLAGS += $(call cc-option,-fstack-usage) > +endif > +CLEAN_FILES += stack-usage > + > # Handle stack protector mode. > ifdef CONFIG_CC_STACKPROTECTOR_REGULAR > stackp-flag := -fstack-protector > @@ -1228,6 +1233,7 @@ help: > echo '' > @echo 'Static analysers' > @echo ' checkstack - Generate a list of stack hogs' > + @echo ' stack-usage - Alternative list of stack hogs' > @echo ' namespacecheck - Name space analysis on compiled kernel' > @echo ' versioncheck - Sanity check on version.h usage' > @echo ' includecheck - Check for duplicate included header files' The other "check" targets have check in their name. Can we follow this naming convention and call it checkstackusage? > @@ -1376,7 +1382,7 @@ clean: $(clean-dirs) > $(call cmd,rmfiles) > @find $(if $(KBUILD_EXTMOD), $(KBUILD_EXTMOD), .) $(RCS_FIND_IGNORE) \ > \( -name '*.[oas]' -o -name '*.ko' -o -name '.*.cmd' \ > - -o -name '*.ko.*' \ > + -o -name '*.ko.*' -o -name '*.su' \ > -o -name '.*.d' -o -name '.*.tmp' -o -name '*.mod.c' \ > -o -name '*.symtypes' -o -name 'modules.order' \ > -o -name modules.builtin -o -name '.tmp_*.o.*' \ > @@ -1432,6 +1438,21 @@ checkstack: > $(OBJDUMP) -d vmlinux $$(find . -name '*.ko') | \ > $(PERL) $(src)/scripts/checkstack.pl $(CHECKSTACK_ARCH) > > +quiet_cmd_cc_stack_usage = SORTING $@ > +cmd_cc_stack_usage = find -name '*.su' | xargs cat | sort -k 2 -n -r > $@ > + > +stack-usage: FORCE > +ifndef CONFIG_CC_STACK_USAGE > + $(Q)echo >&2 'Please rebuild with CONFIG_CC_STACK_USAGE=y' && false > +endif > + $(call cmd,cc_stack_usage) > + $(Q)echo '' > + $(Q)echo 'Unbounded dynamic stack frames:' > + $(Q)grep '\bdynamic$$' $@ || true > + $(Q)echo '' > + $(Q)echo 'Top 100 stack frames:' > + $(Q)head -100 < $@ > + All this belongs to a script - for example named scripts/checkstackusage.sh The top-level Makefile shall be a thin driver and there are too much functionality in that file. but this is no excuse for adding more. Sam -- To unsubscribe from this list: send the line "unsubscribe linux-kbuild" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Sun, Jul 13, 2014 at 1:31 PM, Sam Ravnborg <sam@ravnborg.org> wrote: > Hi Konstantin. > > On Sun, Jul 13, 2014 at 12:24:50AM +0400, Konstantin Khlebnikov wrote: >> This patch adds option for collecting stack usage statistics >> via gcc option '-fstack-usage' (needs gcc 4.6 or newer). >> For each .o file gcc dumps stack-frame sizes into .su file. >> >> File format: >> <file>:<line>:<column>:<function> <size> <qualifiers> >> >> select.c:870:5:do_sys_poll 1040 static >> rtnetlink.c:1880:12:rtnl_newlink 552 dynamic >> random.c:1334:1:random_read 664 dynamic,bounded >> >> Quote from the gcc manpage about qualifiers: >> >> The qualifier "static" means that the function >> manipulates the stack statically: a fixed number of bytes >> are allocated for the frame on function entry and >> released on function exit; no stack adjustments are >> otherwise made in the function. The second field is this >> fixed number of bytes. >> >> The qualifier "dynamic" means that the function >> manipulates the stack dynamically: in addition to the >> static allocation described above, stack adjustments are >> made in the body of the function, for example to push/pop >> arguments around function calls. If the qualifier >> "bounded" is also present, the amount of these >> adjustments is bounded at compile time and the second >> field is an upper bound of the total amount of stack used >> by the function. If it is not present, the amount of >> these adjustments is not bounded at compile time and the >> second field only represents the bounded part. >> >> In comparison to the scripts/checkstack.pl this method is more accurate. >> It takes into account function arguments, push/pop pairs, frame >> pointers and saved return address. It gives information about all >> functions except purely assember (for example sha1_transform_avx2). >> >> Also this patch adds make target 'stack-usage' which combines all .su >> files into one sorted 'stack-usage'. Also prints top 100 and all unbounded >> dynamic stack frames. >> >> Signed-off-by: Konstantin Khlebnikov <koct9i@gmail.com> >> --- >> .gitignore | 3 +++ >> Makefile | 23 ++++++++++++++++++++++- >> lib/Kconfig.debug | 9 +++++++++ >> 3 files changed, 34 insertions(+), 1 deletion(-) >> >> diff --git a/.gitignore b/.gitignore >> index f4c0b09..96b8984 100644 >> --- a/.gitignore >> +++ b/.gitignore >> @@ -85,6 +85,9 @@ GTAGS >> *~ >> \#*# >> >> +*.su >> +/stack-usage >> + >> # >> # Leavings from module signing >> # >> diff --git a/Makefile b/Makefile >> index 2167084..29fe588 100644 >> --- a/Makefile >> +++ b/Makefile >> @@ -635,6 +635,11 @@ ifneq ($(CONFIG_FRAME_WARN),0) >> KBUILD_CFLAGS += $(call cc-option,-Wframe-larger-than=${CONFIG_FRAME_WARN}) >> endif >> >> +ifdef CONFIG_CC_STACK_USAGE >> +KBUILD_CFLAGS += $(call cc-option,-fstack-usage) >> +endif >> +CLEAN_FILES += stack-usage >> + >> # Handle stack protector mode. >> ifdef CONFIG_CC_STACKPROTECTOR_REGULAR >> stackp-flag := -fstack-protector >> @@ -1228,6 +1233,7 @@ help: >> echo '' >> @echo 'Static analysers' >> @echo ' checkstack - Generate a list of stack hogs' >> + @echo ' stack-usage - Alternative list of stack hogs' >> @echo ' namespacecheck - Name space analysis on compiled kernel' >> @echo ' versioncheck - Sanity check on version.h usage' >> @echo ' includecheck - Check for duplicate included header files' > The other "check" targets have check in their name. > Can we follow this naming convention and call it checkstackusage? Most of them, not all of them. export_report and headerdep dont have it. And it's not just a check it generates file as result. So, since I haven't found better name I decided to use the same words as in the gcc option. > > >> @@ -1376,7 +1382,7 @@ clean: $(clean-dirs) >> $(call cmd,rmfiles) >> @find $(if $(KBUILD_EXTMOD), $(KBUILD_EXTMOD), .) $(RCS_FIND_IGNORE) \ >> \( -name '*.[oas]' -o -name '*.ko' -o -name '.*.cmd' \ >> - -o -name '*.ko.*' \ >> + -o -name '*.ko.*' -o -name '*.su' \ >> -o -name '.*.d' -o -name '.*.tmp' -o -name '*.mod.c' \ >> -o -name '*.symtypes' -o -name 'modules.order' \ >> -o -name modules.builtin -o -name '.tmp_*.o.*' \ >> @@ -1432,6 +1438,21 @@ checkstack: >> $(OBJDUMP) -d vmlinux $$(find . -name '*.ko') | \ >> $(PERL) $(src)/scripts/checkstack.pl $(CHECKSTACK_ARCH) >> >> +quiet_cmd_cc_stack_usage = SORTING $@ >> +cmd_cc_stack_usage = find -name '*.su' | xargs cat | sort -k 2 -n -r > $@ >> + >> +stack-usage: FORCE >> +ifndef CONFIG_CC_STACK_USAGE >> + $(Q)echo >&2 'Please rebuild with CONFIG_CC_STACK_USAGE=y' && false >> +endif >> + $(call cmd,cc_stack_usage) >> + $(Q)echo '' >> + $(Q)echo 'Unbounded dynamic stack frames:' >> + $(Q)grep '\bdynamic$$' $@ || true >> + $(Q)echo '' >> + $(Q)echo 'Top 100 stack frames:' >> + $(Q)head -100 < $@ >> + > > All this belongs to a script - for example named > > scripts/checkstackusage.sh > > The top-level Makefile shall be a thin driver and there are too much functionality > in that file. but this is no excuse for adding more. Ok, I'll think about it. > > Sam -- To unsubscribe from this list: send the line "unsubscribe linux-kbuild" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/.gitignore b/.gitignore index f4c0b09..96b8984 100644 --- a/.gitignore +++ b/.gitignore @@ -85,6 +85,9 @@ GTAGS *~ \#*# +*.su +/stack-usage + # # Leavings from module signing # diff --git a/Makefile b/Makefile index 2167084..29fe588 100644 --- a/Makefile +++ b/Makefile @@ -635,6 +635,11 @@ ifneq ($(CONFIG_FRAME_WARN),0) KBUILD_CFLAGS += $(call cc-option,-Wframe-larger-than=${CONFIG_FRAME_WARN}) endif +ifdef CONFIG_CC_STACK_USAGE +KBUILD_CFLAGS += $(call cc-option,-fstack-usage) +endif +CLEAN_FILES += stack-usage + # Handle stack protector mode. ifdef CONFIG_CC_STACKPROTECTOR_REGULAR stackp-flag := -fstack-protector @@ -1228,6 +1233,7 @@ help: echo '' @echo 'Static analysers' @echo ' checkstack - Generate a list of stack hogs' + @echo ' stack-usage - Alternative list of stack hogs' @echo ' namespacecheck - Name space analysis on compiled kernel' @echo ' versioncheck - Sanity check on version.h usage' @echo ' includecheck - Check for duplicate included header files' @@ -1376,7 +1382,7 @@ clean: $(clean-dirs) $(call cmd,rmfiles) @find $(if $(KBUILD_EXTMOD), $(KBUILD_EXTMOD), .) $(RCS_FIND_IGNORE) \ \( -name '*.[oas]' -o -name '*.ko' -o -name '.*.cmd' \ - -o -name '*.ko.*' \ + -o -name '*.ko.*' -o -name '*.su' \ -o -name '.*.d' -o -name '.*.tmp' -o -name '*.mod.c' \ -o -name '*.symtypes' -o -name 'modules.order' \ -o -name modules.builtin -o -name '.tmp_*.o.*' \ @@ -1432,6 +1438,21 @@ checkstack: $(OBJDUMP) -d vmlinux $$(find . -name '*.ko') | \ $(PERL) $(src)/scripts/checkstack.pl $(CHECKSTACK_ARCH) +quiet_cmd_cc_stack_usage = SORTING $@ +cmd_cc_stack_usage = find -name '*.su' | xargs cat | sort -k 2 -n -r > $@ + +stack-usage: FORCE +ifndef CONFIG_CC_STACK_USAGE + $(Q)echo >&2 'Please rebuild with CONFIG_CC_STACK_USAGE=y' && false +endif + $(call cmd,cc_stack_usage) + $(Q)echo '' + $(Q)echo 'Unbounded dynamic stack frames:' + $(Q)grep '\bdynamic$$' $@ || true + $(Q)echo '' + $(Q)echo 'Top 100 stack frames:' + $(Q)head -100 < $@ + kernelrelease: @echo "$(KERNELVERSION)$$($(CONFIG_SHELL) $(srctree)/scripts/setlocalversion $(srctree))" diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug index 7a638aa..3b30853 100644 --- a/lib/Kconfig.debug +++ b/lib/Kconfig.debug @@ -170,6 +170,15 @@ config FRAME_WARN Setting it to 0 disables the warning. Requires gcc 4.4 +config CC_STACK_USAGE + bool "Collect statistics about stack usage (needs gcc 4.6)" + help + This enables gcc option -fstack-usage which saves information + about stack frames in per-function basis into '.su' files. + + Run 'make stack-usage' after building kernel and modules to + combine all information in one file. + config STRIP_ASM_SYMS bool "Strip assembler-generated symbols during link" default n
This patch adds option for collecting stack usage statistics via gcc option '-fstack-usage' (needs gcc 4.6 or newer). For each .o file gcc dumps stack-frame sizes into .su file. File format: <file>:<line>:<column>:<function> <size> <qualifiers> select.c:870:5:do_sys_poll 1040 static rtnetlink.c:1880:12:rtnl_newlink 552 dynamic random.c:1334:1:random_read 664 dynamic,bounded Quote from the gcc manpage about qualifiers: The qualifier "static" means that the function manipulates the stack statically: a fixed number of bytes are allocated for the frame on function entry and released on function exit; no stack adjustments are otherwise made in the function. The second field is this fixed number of bytes. The qualifier "dynamic" means that the function manipulates the stack dynamically: in addition to the static allocation described above, stack adjustments are made in the body of the function, for example to push/pop arguments around function calls. If the qualifier "bounded" is also present, the amount of these adjustments is bounded at compile time and the second field is an upper bound of the total amount of stack used by the function. If it is not present, the amount of these adjustments is not bounded at compile time and the second field only represents the bounded part. In comparison to the scripts/checkstack.pl this method is more accurate. It takes into account function arguments, push/pop pairs, frame pointers and saved return address. It gives information about all functions except purely assember (for example sha1_transform_avx2). Also this patch adds make target 'stack-usage' which combines all .su files into one sorted 'stack-usage'. Also prints top 100 and all unbounded dynamic stack frames. Signed-off-by: Konstantin Khlebnikov <koct9i@gmail.com> --- .gitignore | 3 +++ Makefile | 23 ++++++++++++++++++++++- lib/Kconfig.debug | 9 +++++++++ 3 files changed, 34 insertions(+), 1 deletion(-) -- To unsubscribe from this list: send the line "unsubscribe linux-kbuild" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html