| Message ID | 20241102100452.793970-1-flo@geekplace.eu (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | kbuild,bpf: pass make jobs' value to pahole | expand |
On 2024-11-02 11:04, Florian Schmaus wrote: > Pass the value of make's -j/--jobs argument to pahole, to avoid out of > memory errors and make pahole respect the "jobs" value of make. > > On systems with little memory but many cores, invoking pahole using -j > without argument potentially creates too many pahole instances, > causing an out-of-memory situation. Instead, we should pass make's > "jobs" value as an argument to pahole's -j, which is likely configured > to be (much) lower than the actual core count on such systems. > > If make was invoked without -j, either via cmdline or MAKEFLAGS, then > JOBS will be simply empty, resulting in the existing behavior, as > expected. > > Signed-off-by: Florian Schmaus <flo@geekplace.eu> As discussed on IRC: Tested-by: Holger Hoffstätte <holger@applied-asynchrony.com> Thanks! Holger
On Sun, Nov 3, 2024 at 9:04 PM Holger Hoffstätte <holger@applied-asynchrony.com> wrote: > > On 2024-11-02 11:04, Florian Schmaus wrote: > > Pass the value of make's -j/--jobs argument to pahole, to avoid out of > > memory errors and make pahole respect the "jobs" value of make. > > > > On systems with little memory but many cores, invoking pahole using -j > > without argument potentially creates too many pahole instances, > > causing an out-of-memory situation. Instead, we should pass make's > > "jobs" value as an argument to pahole's -j, which is likely configured > > to be (much) lower than the actual core count on such systems. > > > > If make was invoked without -j, either via cmdline or MAKEFLAGS, then > > JOBS will be simply empty, resulting in the existing behavior, as > > expected. > > > > Signed-off-by: Florian Schmaus <flo@geekplace.eu> > > As discussed on IRC: Do not do this. Others do not see what was discussed. I guess the right thing to do is to join the jobserver. https://www.gnu.org/software/make/manual/html_node/POSIX-Jobserver.html
On 03/11/2024 14.22, Masahiro Yamada wrote: > On Sun, Nov 3, 2024 at 9:04 PM Holger Hoffstätte > <holger@applied-asynchrony.com> wrote: >> >> On 2024-11-02 11:04, Florian Schmaus wrote: >>> Pass the value of make's -j/--jobs argument to pahole, to avoid out of >>> memory errors and make pahole respect the "jobs" value of make. >>> >>> On systems with little memory but many cores, invoking pahole using -j >>> without argument potentially creates too many pahole instances, >>> causing an out-of-memory situation. Instead, we should pass make's >>> "jobs" value as an argument to pahole's -j, which is likely configured >>> to be (much) lower than the actual core count on such systems. >>> >>> If make was invoked without -j, either via cmdline or MAKEFLAGS, then >>> JOBS will be simply empty, resulting in the existing behavior, as >>> expected. >>> >>> Signed-off-by: Florian Schmaus <flo@geekplace.eu> >> >> As discussed on IRC: > > Do not do this. Others do not see what was discussed. Sorry, you are right. However, not much was discussed. Holger just pointed out that the memory usage of pahole was already reported as problematic in https://lore.kernel.org/lkml/20240820085950.200358-1-jirislaby@kernel.org/ My patch would potentially help there as well, as it allows the user to limit the number of threads used by pahole. > I guess the right thing to do is to join the jobserver. > > https://www.gnu.org/software/make/manual/html_node/POSIX-Jobserver.html Yes, this would be the ideal solution. Until it is implemented, the proposed patch is probably the next best thing. - Florian
On 04/11/2024 12:52, Florian Schmaus wrote: > On 03/11/2024 14.22, Masahiro Yamada wrote: >> On Sun, Nov 3, 2024 at 9:04 PM Holger Hoffstätte >> <holger@applied-asynchrony.com> wrote: >>> >>> On 2024-11-02 11:04, Florian Schmaus wrote: >>>> Pass the value of make's -j/--jobs argument to pahole, to avoid out of >>>> memory errors and make pahole respect the "jobs" value of make. >>>> >>>> On systems with little memory but many cores, invoking pahole using -j >>>> without argument potentially creates too many pahole instances, >>>> causing an out-of-memory situation. Instead, we should pass make's >>>> "jobs" value as an argument to pahole's -j, which is likely configured >>>> to be (much) lower than the actual core count on such systems. >>>> >>>> If make was invoked without -j, either via cmdline or MAKEFLAGS, then >>>> JOBS will be simply empty, resulting in the existing behavior, as >>>> expected. >>>> >>>> Signed-off-by: Florian Schmaus <flo@geekplace.eu> >>> >>> As discussed on IRC: >> >> Do not do this. Others do not see what was discussed. > > Sorry, you are right. However, not much was discussed. Holger just > pointed out that the memory usage of pahole was already reported as > problematic in > > https://lore.kernel.org/lkml/20240820085950.200358-1-jirislaby@kernel.org/ > > My patch would potentially help there as well, as it allows the user to > limit the number of threads used by pahole. > > >> I guess the right thing to do is to join the jobserver. >> >> https://www.gnu.org/software/make/manual/html_node/POSIX-Jobserver.html > > Yes, this would be the ideal solution. Until it is implemented, the > proposed patch is probably the next best thing. > > - Florian > If you haven't already, I'd suggest testing the latest pahole (building from master branch). Significant improvements in memory utilization have been made and more are planned post the 1.28 release. In terms of the patch itself, respecting the "make -j" value seems right to me. Thanks! Alan
Hello: This patch was applied to bpf/bpf-next.git (master) by Andrii Nakryiko <andrii@kernel.org>: On Sat, 2 Nov 2024 11:04:51 +0100 you wrote: > Pass the value of make's -j/--jobs argument to pahole, to avoid out of > memory errors and make pahole respect the "jobs" value of make. > > On systems with little memory but many cores, invoking pahole using -j > without argument potentially creates too many pahole instances, > causing an out-of-memory situation. Instead, we should pass make's > "jobs" value as an argument to pahole's -j, which is likely configured > to be (much) lower than the actual core count on such systems. > > [...] Here is the summary with links: - kbuild,bpf: pass make jobs' value to pahole https://git.kernel.org/bpf/bpf-next/c/09048d22b782 You are awesome, thank you!
diff --git a/scripts/Makefile.btf b/scripts/Makefile.btf index b75f09f3f424..c3cbeb13de50 100644 --- a/scripts/Makefile.btf +++ b/scripts/Makefile.btf @@ -3,6 +3,8 @@ pahole-ver := $(CONFIG_PAHOLE_VERSION) pahole-flags-y := +JOBS := $(patsubst -j%,%,$(filter -j%,$(MAKEFLAGS))) + ifeq ($(call test-le, $(pahole-ver), 125),y) # pahole 1.18 through 1.21 can't handle zero-sized per-CPU vars @@ -12,14 +14,14 @@ endif pahole-flags-$(call test-ge, $(pahole-ver), 121) += --btf_gen_floats -pahole-flags-$(call test-ge, $(pahole-ver), 122) += -j +pahole-flags-$(call test-ge, $(pahole-ver), 122) += -j$(JOBS) pahole-flags-$(call test-ge, $(pahole-ver), 125) += --skip_encoding_btf_inconsistent_proto --btf_gen_optimized else # Switch to using --btf_features for v1.26 and later. -pahole-flags-$(call test-ge, $(pahole-ver), 126) = -j --btf_features=encode_force,var,float,enum64,decl_tag,type_tag,optimized_func,consistent_func,decl_tag_kfuncs +pahole-flags-$(call test-ge, $(pahole-ver), 126) = -j$(JOBS) --btf_features=encode_force,var,float,enum64,decl_tag,type_tag,optimized_func,consistent_func,decl_tag_kfuncs ifneq ($(KBUILD_EXTMOD),) module-pahole-flags-$(call test-ge, $(pahole-ver), 126) += --btf_features=distilled_base
Pass the value of make's -j/--jobs argument to pahole, to avoid out of memory errors and make pahole respect the "jobs" value of make. On systems with little memory but many cores, invoking pahole using -j without argument potentially creates too many pahole instances, causing an out-of-memory situation. Instead, we should pass make's "jobs" value as an argument to pahole's -j, which is likely configured to be (much) lower than the actual core count on such systems. If make was invoked without -j, either via cmdline or MAKEFLAGS, then JOBS will be simply empty, resulting in the existing behavior, as expected. Signed-off-by: Florian Schmaus <flo@geekplace.eu> --- scripts/Makefile.btf | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)