* Re: [PATCH] tools/build: make in-target rule robust against too long argument error
2026-03-03 22:33 ` [PATCH] tools/build: make in-target rule robust against too long argument error Ian Rogers
@ 2026-03-04 2:35 ` Changqing Li
2026-03-04 14:13 ` Arnaldo Carvalho de Melo
1 sibling, 0 replies; 3+ messages in thread
From: Changqing Li @ 2026-03-04 2:35 UTC (permalink / raw)
To: Ian Rogers, Arnaldo Carvalho de Melo, Namhyung Kim
Cc: namhyung, james.clark, charlie, linux-kernel, Markus Mayer,
linux-perf-users
On 3/4/26 06:33, Ian Rogers wrote:
> CAUTION: This email comes from a non Wind River email account!
> Do not click links or open attachments unless you recognize the sender and know the content is safe.
>
> On Mon, Oct 27, 2025 at 7:16 PM Changqing Li <changqing.li@windriver.com> wrote:
>> Hi, Ian
>>
>> Kindly ping. Could you reproduce it? Any comments?
> Sorry Changqing,
>
> I was reminded by:
> https://lore.kernel.org/lkml/20260303211503.165337-1-mmayer@broadcom.com/
> that "argument list too long" is a problem for others.
>
> I tried reproducing it, but I think it doesn't fail for me, possibly
> because MAX_ARGS is larger on my system:
> ```
> $ xargs --show-limits
> Your environment variables take up 6133 bytes
> POSIX upper limit on argument length (this system): 2088971
> POSIX smallest allowable upper limit on argument length (all systems): 4096
> Maximum length of command we could actually use: 2082838
> Size of command buffer we are actually using: 131072
> Maximum parallelism (--max-procs must be no greater): 2147483647
> ```
>
> Anyway, the change makes sense to me.
>
> Reviewed-by: Ian Rogers <irogers@google.com>
>
> Thanks,
> Ian
>
Hi, Ian
Thanks. I also cannot reproduce with previous length of tmpdir, maybe
caused by perf source changes during this period，
but I can still reproduce after I make the tmpdir longer.
$ xargs --show-limits
Your environment variables take up 3124 bytes
POSIX upper limit on argument length (this system): 2091980
POSIX smallest allowable upper limit on argument length (all systems): 4096
Maximum length of command we could actually use: 2088856
Size of command buffer we are actually using: 131072
Maximum parallelism (--max-procs must be no greater): 2147483647
Regards
Changqing
>> Regards
>>
>> Changqing
>>
>> On 8/19/25 11:01, Changqing Li wrote:
>>> On 8/19/25 04:38, Ian Rogers wrote:
>>>> CAUTION: This email comes from a non Wind River email account!
>>>> Do not click links or open attachments unless you recognize the
>>>> sender and know the content is safe.
>>>>
>>>> On Mon, Jul 28, 2025 at 2:31 AM <changqing.li@windriver.com> wrote:
>>>>> From: Changqing Li <changqing.li@windriver.com>
>>>>>
>>>>> The command length of in-target scales with the depth of the directory
>>>>> times the number of objects in the Makefile. When there are many
>>>>> objects, and O=[absolute_path] is set, and the absolute_path is
>>>>> relatively long. It is possile that this line "$(call
>>>>> if_changed,$(host)ld_multi)" will report error:
>>>>> "make[4]: /bin/sh: Argument list too long"
>>>>>
>>>>> For example, build perf tools with O=/long/output/path
>>>>>
>>>>> Like built-in.a and *.mod rules in scripts/Makefile.build, add
>>>>> $(objpredix)/ by the shell command instead of by Make's builtin
>>>>> function.
>>>>>
>>>>> Signed-off-by: Changqing Li <changqing.li@windriver.com>
>>>> Thanks Changqing, the change makes sense to me. The printf is pushing
>>>> the values into xargs rather than using $^ with ld. I've tried
>>>> reproducing the error to test your fix by creating long directory
>>>> names in /tmp and then passing them to O=. I've not been able to do
>>>> this. Could you send a reproduction for me to test?
>>>>
>>>> Thanks,
>>>> Ian
>>> Hi, Ian
>>>
>>> Thanks. Here is my steps:
>>>
>>> 1. git clone https://github.com/torvalds/linux.git
>>>
>>> 2. cd linux/tools/perf/
>>>
>>> 3. mkdir
>>> /tmp/01234567890123456789/0123456789012345678901234567890123456789/0123456789012345678901234567890123456789/012345678901234567890123456789/01234567890123456789/012345678901234567890123456789/01234567890123456789012345678/9012345678901234567890123456789/012345678901234567890123456789/012345678901234567890123456789/01234567890123456789012345678901234567890123456789012345678901234567890123/perf-1.0
>>>
>>> 4. make NO_LIBTRACEEVENT=1
>>> O=/tmp/01234567890123456789/0123456789012345678901234567890123456789/0123456789012345678901234567890123456789/012345678901234567890123456789/01234567890123456789/012345678901234567890123456789/01234567890123456789012345678/9012345678901234567890123456789/012345678901234567890123456789/012345678901234567890123456789/01234567890123456789012345678901234567890123456789012345678901234567890123/perf-1.0
>>>
>>>
>>> 5. Above step failed with error:
>>>
>>> make[4]: /bin/sh: Argument list too long
>>>
>>> make[4]: *** [/buildarea2/cli10/tmp/linux/tools/build/Makefile.build
>>> <https://urldefense.com/v3/__http://Makefile.build__;!!AjveYdw8EvQ!c2n7zPX4Vrhwd7a9UxsjCXBGO5V8Llmon1egvupz4kHL4869ilesQgRgQZO8nj8AWwQJUlY9DOoWPD688quA73jNugR6IZRxy09Eig$>:148:
>>> /tmp/01234567890123456789/0123456789012345678901234567890123456789/0123456789012345678901234567890123456789/012345678901234567890123456789/01234567890123456789/012345678901234567890123456789/01234567890123456789012345678/9012345678901234567890123456789/012345678901234567890123456789/012345678901234567890123456789/01234567890123456789012345678901234567890123456789012345678901234567890123/perf-1.0/util/perf-util-in.o]
>>> Error 127
>>>
>>> make[3]: *** [/buildarea2/cli10/tmp/linux/tools/build/Makefile.build
>>> <https://urldefense.com/v3/__http://Makefile.build__;!!AjveYdw8EvQ!c2n7zPX4Vrhwd7a9UxsjCXBGO5V8Llmon1egvupz4kHL4869ilesQgRgQZO8nj8AWwQJUlY9DOoWPD688quA73jNugR6IZRxy09Eig$>:142:
>>> util] Error 2
>>>
>>> make[2]: *** [Makefile.perf:797:
>>> /tmp/01234567890123456789/0123456789012345678901234567890123456789/0123456789012345678901234567890123456789/012345678901234567890123456789/01234567890123456789/012345678901234567890123456789/01234567890123456789012345678/9012345678901234567890123456789/012345678901234567890123456789/012345678901234567890123456789/01234567890123456789012345678901234567890123456789012345678901234567890123/perf-1.0/perf-util-in.o]
>>> Error 2
>>>
>>>
>>> Regards
>>>
>>> Changqing
>>>>> ---
>>>>> tools/build/Makefile.build | 6 ++++--
>>>>> 1 file changed, 4 insertions(+), 2 deletions(-)
>>>>>
>>>>> diff --git a/tools/build/Makefile.build b/tools/build/Makefile.build
>>>>> index 3584ff308607..39066a3ef2fc 100644
>>>>> --- a/tools/build/Makefile.build
>>>>> +++ b/tools/build/Makefile.build
>>>>> @@ -70,11 +70,13 @@ quiet_cmd_gen = GEN $@
>>>>> # If there's nothing to link, create empty $@ object.
>>>>> quiet_cmd_ld_multi = LD $@
>>>>> cmd_ld_multi = $(if $(strip $(obj-y)),\
>>>>> - $(LD) -r -o $@ $(filter $(obj-y),$^),rm -f
>>>>> $@; $(AR) rcs $@)
>>>>> + printf "$(objprefix)%s " $(patsubst
>>>>> $(objprefix)%,%,$(filter $(obj-y),$^)) | \
>>>>> + xargs $(LD) -r -o $@,rm -f $@; $(AR) rcs $@)
>>>>>
>>>>> quiet_cmd_host_ld_multi = HOSTLD $@
>>>>> cmd_host_ld_multi = $(if $(strip $(obj-y)),\
>>>>> - $(HOSTLD) -r -o $@ $(filter
>>>>> $(obj-y),$^),rm -f $@; $(HOSTAR) rcs $@)
>>>>> + printf "$(objprefix)%s " $(patsubst
>>>>> $(objprefix)%,%,$(filter $(obj-y),$^)) | \
>>>>> + xargs $(HOSTLD) -r -o $@,rm -f $@;
>>>>> $(HOSTAR) rcs $@)
>>>>>
>>>>> ifneq ($(filter $(obj),$(hostprogs)),)
>>>>> host = host_
>>>>> --
>>>>> 2.34.1
>>>>>
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] tools/build: make in-target rule robust against too long argument error
2026-03-03 22:33 ` [PATCH] tools/build: make in-target rule robust against too long argument error Ian Rogers
2026-03-04 2:35 ` Changqing Li
@ 2026-03-04 14:13 ` Arnaldo Carvalho de Melo
1 sibling, 0 replies; 3+ messages in thread
From: Arnaldo Carvalho de Melo @ 2026-03-04 14:13 UTC (permalink / raw)
To: Ian Rogers
Cc: Changqing Li, Namhyung Kim, namhyung, james.clark, charlie,
linux-kernel, Markus Mayer, linux-perf-users
On Tue, Mar 03, 2026 at 02:33:04PM -0800, Ian Rogers wrote:
> On Mon, Oct 27, 2025 at 7:16 PM Changqing Li <changqing.li@windriver.com> wrote:
> > Kindly ping. Could you reproduce it? Any comments?
> Sorry Changqing,
> I was reminded by:
> https://lore.kernel.org/lkml/20260303211503.165337-1-mmayer@broadcom.com/
> that "argument list too long" is a problem for others.
> I tried reproducing it, but I think it doesn't fail for me, possibly
> because MAX_ARGS is larger on my system:
> ```
> $ xargs --show-limits
> Your environment variables take up 6133 bytes
> POSIX upper limit on argument length (this system): 2088971
> POSIX smallest allowable upper limit on argument length (all systems): 4096
> Maximum length of command we could actually use: 2082838
> Size of command buffer we are actually using: 131072
> Maximum parallelism (--max-procs must be no greater): 2147483647
> ```
> Anyway, the change makes sense to me.
> Reviewed-by: Ian Rogers <irogers@google.com>
Ok, so I'll pick this one and the other for rm that you mentioned:
https://lore.kernel.org/lkml/20260303211503.165337-1-mmayer@broadcom.com
I'll have it in perf-tools, for v7.0.
- Arnaldo
^ permalink raw reply [flat|nested] 3+ messages in thread