Masahiro,
Thanks, I was not aware of that patch. That does fix the issue for the
most part. A rebuild now consistently does the same work, and does not
unnecessarily update the timestamp of bzImage as before. It still
takes 3 seconds, but maybe that is necessary? I don't know exactly
what it is doing there.
time make -j16 bzImage
CHK include/config/kernel.release
CHK include/generated/uapi/linux/version.h
DESCEND objtool
CHK include/generated/utsrelease.h
CHK scripts/mod/devicetable-offsets.h
CHK include/generated/timeconst.h
CHK include/generated/bounds.h
CHK include/generated/asm-offsets.h
CALL scripts/checksyscalls.sh
CHK include/generated/compile.h
Kernel: arch/x86/boot/bzImage is ready (#1)
real 0m3.027s
user 0m11.183s
sys 0m4.755s
stat --printf 'mtime: %y\n' arch/x86/boot/bzImage
mtime: 2018-08-23 08:04:55.551644635 -0400
time make -j16 bzImage
CHK include/config/kernel.release
CHK include/generated/uapi/linux/version.h
DESCEND objtool
CHK include/generated/utsrelease.h
CHK scripts/mod/devicetable-offsets.h
CHK include/generated/timeconst.h
CHK include/generated/bounds.h
CHK include/generated/asm-offsets.h
CALL scripts/checksyscalls.sh
CHK include/generated/compile.h
Kernel: arch/x86/boot/bzImage is ready (#1)
real 0m2.966s
user 0m10.917s
sys 0m4.834s
stat --printf 'mtime: %y\n' arch/x86/boot/bzImage
mtime: 2018-08-23 08:04:55.551644635 -0400
time make -j16 bzImage
CHK include/config/kernel.release
CHK include/generated/uapi/linux/version.h
DESCEND objtool
CHK include/generated/utsrelease.h
CHK scripts/mod/devicetable-offsets.h
CHK include/generated/timeconst.h
CHK include/generated/bounds.h
CHK include/generated/asm-offsets.h
CALL scripts/checksyscalls.sh
CHK include/generated/compile.h
Kernel: arch/x86/boot/bzImage is ready (#1)
real 0m2.954s
user 0m11.151s
sys 0m4.734s
stat --printf 'mtime: %y\n' arch/x86/boot/bzImage
mtime: 2018-08-23 08:04:55.551644635 -0400
Sincerely,
Casey Fitzpatrick
On Wed, Aug 22, 2018 at 9:54 PM Masahiro Yamada
<yamada.masahiro@xxxxxxxxxxxxx> wrote:
>
> Hi Casey,
>
> 2018-08-23 8:21 GMT+09:00 Casey Fitzpatrick <kcghost@xxxxxxxxx>:
> > I have observed that attempting to rebuild the kernel with no changes
> > to source files:
> > 1. Takes more time than I should hope is necessary; I would hope it
> > would only check timestamps and determine the target does not need
> > updating rather than do work.
> > 2. Switches between two states of determining what work is needed to do.
> > 3. Unnecessarily updates timestamp on bzImage (and maybe others)
> > despite not actually modifying the file.
> >
> > I am working with v4.14.66 building from an Ubuntu 17.10 machine.
> > The following summarized output demonstrates the issue (I believe the
> > same happens with the default target).
> > Notice that despite no changes made to sources, a rebuild takes 3
> > seconds on my machine, does some work and typically updates the
> > timestamp of the target.
> > Also notice that the work is oddly inconsistent, the third time I ran
> > the same build with no changes it did different work. And in that case
> > it does not update the timestamp of the target, which is preferable.
> >
> > make distclean
> > make x86_64_defconfig
> > time make -j16 bzImage
> > ...(lots of output, actually building all the sources)
> > OBJCOPY arch/x86/boot/setup.bin
> > BUILD arch/x86/boot/bzImage
> > Setup is 15548 bytes (padded to 15872 bytes).
> > System is 7789 kB
> > CRC 5b2b5059
> > Kernel: arch/x86/boot/bzImage is ready (#1)
> >
> > real 4m15.147s
> > user 27m21.182s
> > sys 2m34.607s
> > stat --printf 'mtime: %y\n' arch/x86/boot/bzImage
> > mtime: 2018-08-22 18:57:10.000000000 -0400
> > time make -j16 bzImage
> > CHK include/config/kernel.release
> > CHK include/generated/uapi/linux/version.h
> > DESCEND objtool
> > CHK scripts/mod/devicetable-offsets.h
> > CHK include/generated/utsrelease.h
> > CHK include/generated/timeconst.h
> > CHK include/generated/bounds.h
> > CHK include/generated/asm-offsets.h
> > CALL scripts/checksyscalls.sh
> > CHK include/generated/compile.h
> > DATAREL arch/x86/boot/compressed/vmlinux
> > BUILD arch/x86/boot/bzImage
> > Setup is 15548 bytes (padded to 15872 bytes).
> > System is 7789 kB
> > CRC 5b2b5059
> > Kernel: arch/x86/boot/bzImage is ready (#1)
> >
> > real 0m3.273s
> > user 0m11.260s
> > sys 0m4.727s
> > stat --printf 'mtime: %y\n' arch/x86/boot/bzImage
> > mtime: 2018-08-22 19:00:26.587233951 -0400
> > time make -j16 bzImage
> > CHK include/config/kernel.release
> > CHK include/generated/uapi/linux/version.h
> > DESCEND objtool
> > CHK include/generated/utsrelease.h
> > CHK scripts/mod/devicetable-offsets.h
> > CHK include/generated/timeconst.h
> > CHK include/generated/bounds.h
> > CHK include/generated/asm-offsets.h
> > CALL scripts/checksyscalls.sh
> > CHK include/generated/compile.h
> > LD arch/x86/boot/compressed/vmlinux
> > ZOFFSET arch/x86/boot/zoffset.h
> > OBJCOPY arch/x86/boot/vmlinux.bin
> > AS arch/x86/boot/header.o
> > LD arch/x86/boot/setup.elf
> > OBJCOPY arch/x86/boot/setup.bin
> > BUILD arch/x86/boot/bzImage
> > Setup is 15548 bytes (padded to 15872 bytes).
> > System is 7789 kB
> > CRC 5b2b5059
> > Kernel: arch/x86/boot/bzImage is ready (#1)
> >
> > real 0m3.136s
> > user 0m11.411s
> > sys 0m4.775s
> > stat --printf 'mtime: %y\n' arch/x86/boot/bzImage
> > mtime: 2018-08-22 19:01:11.583377706 -0400
> > time make -j16 bzImage
> > CHK include/config/kernel.release
> > CHK include/generated/uapi/linux/version.h
> > DESCEND objtool
> > CHK include/generated/utsrelease.h
> > CHK scripts/mod/devicetable-offsets.h
> > CHK include/generated/timeconst.h
> > CHK include/generated/bounds.h
> > CHK include/generated/asm-offsets.h
> > CALL scripts/checksyscalls.sh
> > CHK include/generated/compile.h
> > DATAREL arch/x86/boot/compressed/vmlinux
> > Kernel: arch/x86/boot/bzImage is ready (#1)
> >
> > real 0m3.118s
> > user 0m11.476s
> > sys 0m4.505s
> > stat --printf 'mtime: %y\n' arch/x86/boot/bzImage
> > mtime: 2018-08-22 19:01:11.583377706 -0400
> >
> > Sincerely,
> > Casey Fitzpatrick
>
>
>
> Could you check if the following commit fixes the problem?
>
>
> commit 92a4728608a8fd228c572bc8ff50dd98aa0ddf2a
> Author: Kees Cook <keescook@xxxxxxxxxxxx>
> Date: Tue Jul 24 16:08:27 2018 -0700
>
> x86/boot: Fix if_changed build flip/flop bug
>
>
>
>
> I am not sure whether it was back-ported to 4.14 kernel or not...
>
>
>
>
>
>
> --
> Best Regards
> Masahiro Yamada
