Re: [PATCH ftrace/core v6 5/5] kselftest, ftrace: Add ftrace IPMODIFY flag test

[Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>]

(CC'ed Shuah, since this is related to kselftest)

(2014/11/24 13:29), Steven Rostedt wrote:
> On Mon, 24 Nov 2014 11:50:06 +0900
> Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> wrote:
> 
>> (2014/11/22 6:03), Steven Rostedt wrote:
>>> On Fri, 21 Nov 2014 05:25:37 -0500
>>> Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> wrote:
>>>
>>>> Add ftrace IPMODIFY flag test to selftest/ftrace. The
>>>> test checks simple ftrace handler insertion and
>>>> combinations of ftrace, kprobe, and jprobe.
>>>> This test requires kernel build tree to build a test
>>>> kernel module and root privilege to run.
>>>>
>>>> Signed-off-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
>>>>
>>>> ---
>>>
>>> Can you make it so that make O=/path/to/build works?
>>
>> Hmm, I always use "O=builddir" option in my build script...
> 
> But do you use O=builddir where builddir is where the kernel was built?
> I don't.

Hmm, yes, I always use kbin/linux-3.<arch>/ for the builddir of
ksrc/linux-3/ source tree.

> 
> Remember, I build on a server and install on a test box. The server
> directory is read only. I can not have O= be the path to where the
> kernel was built. I have to use a different directory.

Oh, I see. So your kernel is built on a directory which mounted as
readonly, and installed on the testbox.

>>> I tried running this on my test box, which has /home/rostedt mounted
>>> read only from my build box, and I get this:

Hmm, btw, would you mount it without "noexec" option?
I think only "ro" option doesn't prohibit executing scripts on it.

>>
>> My case is just to keep the source tree clean, so source tree is
>> still writable and builddir is under another directory in /home/<user>/.
>>
>>
>>> [root@bxtest ftrace]# make O=/tmp
>>> for TARGET in ipmodify; do \
>>> 	make -C $TARGET; \
>>> done;
>>> make[1]: Entering directory `/work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify'
>>> make -C /lib/modules/3.18.0-rc1-test+/build M=/work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify ipmodify.ko
>>> make[2]: Entering directory `/home/rostedt/work/git/nobackup/bxtest/trace'
>>>   CC [M]  /work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify/ipmodify.o
>>> /bin/sh: ./scripts/recordmcount: Permission denied
>>
>> Hmm, this error looks odd, not ENOENT but EPERM??
>> Could you run it again with V=1 ?
> 
> Sure, here's the output:
> 
> [root@bxtest ftrace]# make V=1 O=/tmp
> for TARGET in ipmodify; do \
> 	make -C $TARGET; \
> done;
> make[1]: Entering directory `/work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify'
> make -C /lib/modules/3.18.0-rc1-test+/build M=/work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify ipmodify.ko
> make[2]: Entering directory `/home/rostedt/work/git/nobackup/bxtest/trace'
> make -C /home/rostedt/work/git/linux-trace.git O=/home/rostedt/work/git/nobackup/bxtest/trace/. ipmodify.ko
> make -C /home/rostedt/work/git/nobackup/bxtest/trace KBUILD_SRC=/home/rostedt/work/git/linux-trace.git \
> -f /home/rostedt/work/git/linux-trace.git/Makefile ipmodify.ko
> test -e include/generated/autoconf.h -a -e include/config/auto.conf || (		\
> echo >&2;							\
> echo >&2 "  ERROR: Kernel configuration is invalid.";		\
> echo >&2 "         include/generated/autoconf.h or include/config/auto.conf are missing.";\
> echo >&2 "         Run 'make oldconfig && make prepare' on kernel src to fix it.";	\
> echo >&2 ;							\
> /bin/false)
> mkdir -p /work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify/.tmp_versions ; rm -f /work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify/.tmp_versions/*
> make KBUILD_MODULES=1   \
> -f /home/rostedt/work/git/linux-trace.git/scripts/Makefile.build obj=/work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify ipmodify.o
>   gcc -Wp,-MD,/work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify/.ipmodify.o.d  -nostdinc -isystem /usr/lib/gcc/x86_64-redhat-linux/4.7.2/include -I/home/rostedt/work/git/linux-trace.git/arch/x86/include -Iarch/x86/include/generated  -I/home/rostedt/work/git/linux-trace.git/include -Iinclude -I/home/rostedt/work/git/linux-trace.git/arch/x86/include/uapi -Iarch/x86/include/generated/uapi -I/home/rostedt/work/git/linux-trace.git/include/uapi -Iinclude/generated/uapi -include /home/rostedt/work/git/linux-trace.git/include/linux/kconfig.h   -I/work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify -D__KERNEL__ -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -Werror-implicit-function-declaration -Wno-format-security -m64 -mno-80387 -mno-fp-ret-in-387 -mtune=generic -mno-red-zone -mcmodel=kernel -funit-at-a-time -maccumulate-outgoing-
> args -DCONFIG_AS_CFI=1 -DCONFIG_AS_CFI_SIGNAL_FRAME=1 -DCONFIG_AS_CFI_SECTIONS=1 -DCONFIG_A
>  S_FXSAVEQ=1 -DCONFIG_AS_CRC32=1 -DCONFIG_AS_AVX=1 -DCONFIG_AS_AVX2=1 -pipe -Wno-sign-compare -fno-asynchronous-unwind-tables -mno-sse -mno-mmx -mno-sse2 -mno-3dnow -mno-avx -fno-delete-null-pointer-checks -Os -Wno-maybe-uninitialized --param=allow-store-data-races=0 -Wframe-larger-than=2048 -fno-stack-protector -Wno-unused-but-set-variable -fno-omit-frame-pointer -fno-optimize-sibling-calls -fno-var-tracking-assignments -g -pg -mfentry -DCC_USING_FENTRY -Wdeclaration-after-statement -Wno-pointer-sign -fno-strict-overflow -fconserve-stack -Werror=implicit-int -Werror=strict-prototypes -DCC_HAVE_ASM_GOTO  -DMODULE  -D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(ipmodify)"  -D"KBUILD_MODNAME=KBUILD_STR(ipmodify)" -c -o /work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify/ipmodify.o /work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify/ipmodify.c
>   if [ "-pg" = "-pg" ]; then if [ /work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify/ipmodify.o != "scripts/mod/empty.o" ]; then ./scripts/recordmcount  "/work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify/ipmodify.o"; fi; fi;
> /bin/sh: ./scripts/recordmcount: Permission denied
> make[5]: *** [/work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify/ipmodify.o] Error 1

thanks. In my case, I didn't have same error.
---
  if [ "-pg" = "-pg" ]; then if [ /home/mhiramat/ksrc/linux-3/tools/testing/selftests/ftrace/ipmodify/ipmodify.o != "scripts/mod/empty.o" ]; then ./scripts/recordmcount
"/home/mhiramat/ksrc/linux-3/tools/testing/selftests/ftrace/ipmodify/ipmodify.o"; fi; fi;
make -f /home/mhiramat/ksrc/linux-3/scripts/Makefile.modpost
---

However, when I dropped exec bit from recordmcount, I got the same error.
---
$ chmod a-x ~/kbin/linux-3.x86_64/scripts/recordmcount
$ make V=1 O=/tmp
[...]
  if [ "-pg" = "-pg" ]; then if [ /home/mhiramat/ksrc/linux-3/tools/testing/selftests/ftrace/ipmodify/ipmodify.o != "scripts/mod/empty.o" ]; then ./scripts/recordmcount
"/home/mhiramat/ksrc/linux-3/tools/testing/selftests/ftrace/ipmodify/ipmodify.o"; fi; fi;
/bin/sh: ./scripts/recordmcount: Permission denied
make[5]: *** [/home/mhiramat/ksrc/linux-3/tools/testing/selftests/ftrace/ipmodify/ipmodify.o] Error 1
---

So, if your server directory is mounted with noexec, it's an environmental
problem. I guess you can not build any kernel drivers on that testbox, can you?

Anyway, this gives us a good question, "should test binaries be made by
server or client(testbox) environment?" This ipmodify driver is a binary
and it should be built with the kernel binary (by server) I think.
But yes, I missed the Makefile didn't allow that (this always referred
installed running kernel builddir).

I hope to have CONFIG_KSELFTEST_BINARIES for Kconfig, or make prep_kselftest
target to build these binaries with kernel...

Shuah, what would you think about this?

> make[4]: *** [ipmodify.ko] Error 2
> make[3]: *** [sub-make] Error 2
> make[2]: *** [__sub-make] Error 2
> make[2]: Leaving directory `/home/rostedt/work/git/nobackup/bxtest/trace'
> make[1]: *** [ipmodify.ko] Error 2
> make[1]: Leaving directory `/work/git/linux-trace.git/tools/testing/selftests/ftrace/ipmodify'
> make: *** [all] Error 2
> 
> 
> Thanks (on my way to bed),

Thank you :),


-- 
Masami HIRAMATSU
Software Platform Research Dept. Linux Technology Research Center
Hitachi, Ltd., Yokohama Research Laboratory
E-mail: masami.hiramatsu.pt@hitachi.com