Can't kill CONFIG_UNWINDER_ORC with fire or plague.

Regression 4.15 Paul Gortmaker 5 participants 8 messages Dec 26, 2017 – Dec 18, 2018

Subsystems (2)

Kbuild Kconfig

lore

Share & tools

lore

Paul Gortmaker

author

2017-12-26 04:30 UTC #0 lore raw

There is a regression new to 4.15 that happens in a rather common workflow which results in the frustrating situation where the user has clearly disabled UNWINDER_ORC in their ".config" file, and yet they still get the immediate false error saying they need libelf-dev since "CONFIG_UNWINDER_ORC=y" is still set, and hence they can not build.

The regression requires UNWINDER_ORC=y (now the default in commit fc72ae40e303) followed by the user subsequently disabling it, which is common if the user doesn't have libelf-dev and doesn't care about ORC.

This happens because the recently added test in the top level Makefile assumes that ,config data will be mapped into include/config/auto.conf either by the end user running "make oldconfig" or the Makefile itself detecting that "silentoldconfig" should be run. As the simple reproducer below shows, this clearly does not happen as expected.

Note that the test for CONFIG_STACK_VALIDATION and libelf-dev in the same place in the Makefile is broken in the same way, but since it is a warning and not an error, nobody cared about the bogus false positives.

Since there is no way I'm going to debug Makefile stuff on Christmas Day, I figured that I should at least report it intstead. The work around is to save your .config and run a "make distclean" or manually clobber the stale include/config/auto.conf or similar. But a frustrated user wouldn't know that w/o the root cause...

    --------------------- reproducer ------------------------

paul@gw:~/git/linux-head$ git describe 
v4.15-rc5
paul@gw:~/git/linux-head$ make distclean 
paul@gw:~/git/linux-head$ make defconfig
  HOSTCC  scripts/basic/fixdep
  HOSTCC  scripts/kconfig/conf.o
  SHIPPED scripts/kconfig/zconf.tab.c
  SHIPPED scripts/kconfig/zconf.lex.c
  HOSTCC  scripts/kconfig/zconf.tab.o
  HOSTLD  scripts/kconfig/conf
*** Default configuration is based on 'x86_64_defconfig'
#
# configuration written to .config
#
paul@gw:~/git/linux-head$ make
scripts/kconfig/conf  --silentoldconfig Kconfig
Makefile:926: *** "Cannot generate ORC metadata for CONFIG_UNWINDER_ORC=y, please install libelf-dev, libelf-devel or elfutils-libelf-devel".  Stop.
paul@gw:~/git/linux-head$ vi .config  # disable ORC in favour of FRAME_POINTER
paul@gw:~/git/linux-head$ make oldconfig
scripts/kconfig/conf  --oldconfig Kconfig
#
# configuration written to .config
#
paul@gw:~/git/linux-head$ grep UNWINDER .config
# CONFIG_UNWINDER_ORC is not set
CONFIG_UNWINDER_FRAME_POINTER=y
paul@gw:~/git/linux-head$ make
Makefile:926: *** "Cannot generate ORC metadata for CONFIG_UNWINDER_ORC=y, please install libelf-dev, libelf-devel or elfutils-libelf-devel".  Stop.
paul@gw:~/git/linux-head$ ls -l .config include/config/auto.conf
-rw-rw-r-- 1 paul paul 115953 Dec 25 22:48 .config
-rw-rw-r-- 1 paul paul  33069 Dec 25 22:46 include/config/auto.conf
paul@gw:~/git/linux-head$ grep UNWINDER include/config/auto.conf
CONFIG_UNWINDER_ORC=y
paul@gw:~/git/linux-head$

    --------------------- reproducer ------------------------

Randy Dunlap

2017-12-28 19:29 UTC #1 lore raw

Quoted 16 lines · Paul Gortmaker

There is a regression new to 4.15 that happens in a rather common workflow which results in the frustrating situation where the user has clearly disabled UNWINDER_ORC in their ".config" file, and yet they still get the immediate false error saying they need libelf-dev since "CONFIG_UNWINDER_ORC=y" is still set, and hence they can not build.
The regression requires UNWINDER_ORC=y (now the default in commit fc72ae40e303) followed by the user subsequently disabling it, which is common if the user doesn't have libelf-dev and doesn't care about ORC.
This happens because the recently added test in the top level Makefile assumes that ,config data will be mapped into include/config/auto.conf either by the end user running "make oldconfig" or the Makefile itself detecting that "silentoldconfig" should be run. As the simple reproducer below shows, this clearly does not happen as expected.

Quoted 3 lines

Note that the test for CONFIG_STACK_VALIDATION and libelf-dev in the same place in the Makefile is broken in the same way, but since it is a warning and not an error, nobody cared about the bogus false positives.

Yeah, that's odd.

Quoted 6 lines

Since there is no way I'm going to debug Makefile stuff on Christmas Day, I figured that I should at least report it intstead. The work around is to save your .config and run a "make distclean" or manually clobber the stale include/config/auto.conf or similar. But a frustrated user wouldn't know that w/o the root cause...

I have tried to reproduce this multiple times but cannot do so. It remains with:

Quoted 1 line

grep UNWINDER .config

# CONFIG_UNWINDER_ORC is not set
CONFIG_UNWINDER_FRAME_POINTER=y

Quoted 39 lines

    --------------------- reproducer ------------------------

paul@gw:~/git/linux-head$ git describe 
v4.15-rc5
paul@gw:~/git/linux-head$ make distclean 
paul@gw:~/git/linux-head$ make defconfig
  HOSTCC  scripts/basic/fixdep
  HOSTCC  scripts/kconfig/conf.o
  SHIPPED scripts/kconfig/zconf.tab.c
  SHIPPED scripts/kconfig/zconf.lex.c
  HOSTCC  scripts/kconfig/zconf.tab.o
  HOSTLD  scripts/kconfig/conf
*** Default configuration is based on 'x86_64_defconfig'
#
# configuration written to .config
#
paul@gw:~/git/linux-head$ make
scripts/kconfig/conf  --silentoldconfig Kconfig
Makefile:926: *** "Cannot generate ORC metadata for CONFIG_UNWINDER_ORC=y, please install libelf-dev, libelf-devel or elfutils-libelf-devel".  Stop.
paul@gw:~/git/linux-head$ vi .config  # disable ORC in favour of FRAME_POINTER
paul@gw:~/git/linux-head$ make oldconfig
scripts/kconfig/conf  --oldconfig Kconfig
#
# configuration written to .config
#
paul@gw:~/git/linux-head$ grep UNWINDER .config
# CONFIG_UNWINDER_ORC is not set
CONFIG_UNWINDER_FRAME_POINTER=y
paul@gw:~/git/linux-head$ make
Makefile:926: *** "Cannot generate ORC metadata for CONFIG_UNWINDER_ORC=y, please install libelf-dev, libelf-devel or elfutils-libelf-devel".  Stop.
paul@gw:~/git/linux-head$ ls -l .config include/config/auto.conf

(quoted diff) +0 -2

-rw-rw-r-- 1 paul paul 115953 Dec 25 22:48 .config
-rw-rw-r-- 1 paul paul  33069 Dec 25 22:46 include/config/auto.conf

paul@gw:~/git/linux-head$ grep UNWINDER include/config/auto.conf
CONFIG_UNWINDER_ORC=y
paul@gw:~/git/linux-head$

    --------------------- reproducer ------------------------

Signature · 1 line

~Randy

Paul Gortmaker

author

2017-12-29 02:14 UTC #2 lore raw

Quoted 30 lines · [Re: [Regression 4.15] Can't kill CONFIG_UNWINDER_ORC with fire or plague.] On…

Quoted 16 lines · Paul Gortmaker
There is a regression new to 4.15 that happens in a rather common workflow which results in the frustrating situation where the user has clearly disabled UNWINDER_ORC in their ".config" file, and yet they still get the immediate false error saying they need libelf-dev since "CONFIG_UNWINDER_ORC=y" is still set, and hence they can not build.
The regression requires UNWINDER_ORC=y (now the default in commit fc72ae40e303) followed by the user subsequently disabling it, which is common if the user doesn't have libelf-dev and doesn't care about ORC.
This happens because the recently added test in the top level Makefile assumes that ,config data will be mapped into include/config/auto.conf either by the end user running "make oldconfig" or the Makefile itself detecting that "silentoldconfig" should be run. As the simple reproducer below shows, this clearly does not happen as expected.
Quoted 3 lines
Note that the test for CONFIG_STACK_VALIDATION and libelf-dev in the same place in the Makefile is broken in the same way, but since it is a warning and not an error, nobody cared about the bogus false positives.
Yeah, that's odd.
Quoted 6 lines
Since there is no way I'm going to debug Makefile stuff on Christmas Day, I figured that I should at least report it intstead. The work around is to save your .config and run a "make distclean" or manually clobber the stale include/config/auto.conf or similar. But a frustrated user wouldn't know that w/o the root cause...

Quoted 6 lines

I have tried to reproduce this multiple times but cannot do so. It remains with:
Quoted 1 line
grep UNWINDER .config
# CONFIG_UNWINDER_ORC is not set
CONFIG_UNWINDER_FRAME_POINTER=y

As it should. Please re-read the reproducer and the above. The data in the .config is exactly as it should be. The include/config/auto.conf having stale data with ORC still listed as enabled is the issue.

Also note you need a machine with libelf-dev not installed, at the risk of stating the obvious.

Paul.
--

Quoted 44 lines

Quoted 40 lines


    --------------------- reproducer ------------------------

paul@gw:~/git/linux-head$ git describe 
v4.15-rc5
paul@gw:~/git/linux-head$ make distclean 
paul@gw:~/git/linux-head$ make defconfig
  HOSTCC  scripts/basic/fixdep
  HOSTCC  scripts/kconfig/conf.o
  SHIPPED scripts/kconfig/zconf.tab.c
  SHIPPED scripts/kconfig/zconf.lex.c
  HOSTCC  scripts/kconfig/zconf.tab.o
  HOSTLD  scripts/kconfig/conf
*** Default configuration is based on 'x86_64_defconfig'
#
# configuration written to .config
#
paul@gw:~/git/linux-head$ make
scripts/kconfig/conf  --silentoldconfig Kconfig
Makefile:926: *** "Cannot generate ORC metadata for CONFIG_UNWINDER_ORC=y, please install libelf-dev, libelf-devel or elfutils-libelf-devel".  Stop.
paul@gw:~/git/linux-head$ vi .config  # disable ORC in favour of FRAME_POINTER
paul@gw:~/git/linux-head$ make oldconfig
scripts/kconfig/conf  --oldconfig Kconfig
#
# configuration written to .config
#
paul@gw:~/git/linux-head$ grep UNWINDER .config
# CONFIG_UNWINDER_ORC is not set
CONFIG_UNWINDER_FRAME_POINTER=y
paul@gw:~/git/linux-head$ make
Makefile:926: *** "Cannot generate ORC metadata for CONFIG_UNWINDER_ORC=y, please install libelf-dev, libelf-devel or elfutils-libelf-devel".  Stop.
paul@gw:~/git/linux-head$ ls -l .config include/config/auto.conf

(quoted diff) +0 -2

-rw-rw-r-- 1 paul paul 115953 Dec 25 22:48 .config
-rw-rw-r-- 1 paul paul  33069 Dec 25 22:46 include/config/auto.conf

paul@gw:~/git/linux-head$ grep UNWINDER include/config/auto.conf
CONFIG_UNWINDER_ORC=y
paul@gw:~/git/linux-head$

    --------------------- reproducer ------------------------

Signature · 1 line

~Randy

Josh Poimboeuf

2017-12-29 16:47 UTC #3 lore raw

Quoted 10 lines · Paul Gortmaker

Quoted 10 lines
Quoted 10 lines
There is a regression new to 4.15 that happens in a rather common workflow which results in the frustrating situation where the user has clearly disabled UNWINDER_ORC in their ".config" file, and yet they still get the immediate false error saying they need libelf-dev since "CONFIG_UNWINDER_ORC=y" is still set, and hence they can not build.
The regression requires UNWINDER_ORC=y (now the default in commit fc72ae40e303) followed by the user subsequently disabling it, which is common if the user doesn't have libelf-dev and doesn't care about ORC.

Quoted 5 lines

Quoted 5 lines
Quoted 5 lines
This happens because the recently added test in the top level Makefile assumes that ,config data will be mapped into include/config/auto.conf either by the end user running "make oldconfig" or the Makefile itself detecting that "silentoldconfig" should be run. As the simple reproducer below shows, this clearly does not happen as expected.

This seems to be related to a kconfig quirk where only silentoldconfig updates the include/config/auto.conf file. The other config targets (oldconfig, defconfig, etc) don't touch it. It seems intentional, but I have no idea why.

That causes the Makefile to get stale data for 'CONFIG_*' variables when it includes auto.conf. So I don't think this is specific to the ORC check. It seems like it could also cause bugs elsewhere.

The below (ugly) patch fixes it, though I'm not sure this is the best way to do it. We probably need Masahiro or Michal to chime in here.

scripts/kconfig/Makefile +10 -0

index 297c1bf35140..3a0d92040a9c 100644
@@ -21,18 +21,23 @@ unexport CONFIG_
 
 xconfig: $(obj)/qconf
 	$< $(silent) $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 gconfig: $(obj)/gconf
 	$< $(silent) $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 menuconfig: $(obj)/mconf
 	$< $(silent) $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 config: $(obj)/conf
 	$< $(silent) --oldaskconfig $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 nconfig: $(obj)/nconf
 	$< $(silent) $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 silentoldconfig: $(obj)/conf
 	$(Q)mkdir -p include/config include/generated

@@ -85,6 +90,7 @@ PHONY += $(simple-targets)
 
 $(simple-targets): $(obj)/conf
 	$< $(silent) --$@ $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 PHONY += oldnoconfig savedefconfig defconfig

@@ -108,9 +114,11 @@ else
 	$(Q)$(MAKE) -f $(srctree)/Makefile $(KBUILD_DEFCONFIG)
 endif
 endif
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 %_defconfig: $(obj)/conf
 	$(Q)$< $(silent) --defconfig=arch/$(SRCARCH)/configs/$@ $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 configfiles=$(wildcard $(srctree)/kernel/configs/$@ $(srctree)/arch/$(SRCARCH)/configs/$@)

@@ -118,6 +126,7 @@ configfiles=$(wildcard $(srctree)/kernel/configs/$@ $(srctree)/arch/$(SRCARCH)/c
 	$(if $(call configfiles),, $(error No configuration exists for this target on this architecture))
 	$(Q)$(CONFIG_SHELL) $(srctree)/scripts/kconfig/merge_config.sh -m .config $(configfiles)
 	+$(Q)yes "" | $(MAKE) -f $(srctree)/Makefile oldconfig
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 PHONY += kvmconfig
 kvmconfig: kvm_guest.config

@@ -130,6 +139,7 @@ xenconfig: xen.config
 PHONY += tinyconfig
 tinyconfig:
 	$(Q)$(MAKE) -f $(srctree)/Makefile allnoconfig tiny.config
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 # Help text used by make help
 help:

Paul Gortmaker

author

2017-12-29 18:19 UTC #4 lore raw

Quoted 8 lines · [Re: [Regression 4.15] Can't kill CONFIG_UNWINDER_ORC with fire or plague.] On…

This seems to be related to a kconfig quirk where only silentoldconfig updates the include/config/auto.conf file. The other config targets (oldconfig, defconfig, etc) don't touch it. It seems intentional, but I have no idea why.
That causes the Makefile to get stale data for 'CONFIG_*' variables when it includes auto.conf. So I don't think this is specific to the ORC check. It seems like it could also cause bugs elsewhere.

OK, good - you agree with my initial diagnosis of stale auto.conf then. Not sure what Randy was testing when he said he couldn't reproduce it.

Quoted 2 lines

The below (ugly) patch fixes it, though I'm not sure this is the best way to do it. We probably need Masahiro or Michal to chime in here.

Yep, that is why I intentionally put the kbuild folks on the To/Cc of the original report (and ran away screaming at the prospect of debugging Makefiles on xmas day). But with holidays and all, it might not be until early January before they have a chance to reply.

Paul.
--

Quoted 65 lines

scripts/kconfig/Makefile +10 -0

index 297c1bf35140..3a0d92040a9c 100644
@@ -21,18 +21,23 @@ unexport CONFIG_
 
 xconfig: $(obj)/qconf
 	$< $(silent) $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 gconfig: $(obj)/gconf
 	$< $(silent) $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 menuconfig: $(obj)/mconf
 	$< $(silent) $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 config: $(obj)/conf
 	$< $(silent) --oldaskconfig $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 nconfig: $(obj)/nconf
 	$< $(silent) $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 silentoldconfig: $(obj)/conf
 	$(Q)mkdir -p include/config include/generated

@@ -85,6 +90,7 @@ PHONY += $(simple-targets)
 
 $(simple-targets): $(obj)/conf
 	$< $(silent) --$@ $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 PHONY += oldnoconfig savedefconfig defconfig

@@ -108,9 +114,11 @@ else
 	$(Q)$(MAKE) -f $(srctree)/Makefile $(KBUILD_DEFCONFIG)
 endif
 endif
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 %_defconfig: $(obj)/conf
 	$(Q)$< $(silent) --defconfig=arch/$(SRCARCH)/configs/$@ $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 configfiles=$(wildcard $(srctree)/kernel/configs/$@ $(srctree)/arch/$(SRCARCH)/configs/$@)

@@ -118,6 +126,7 @@ configfiles=$(wildcard $(srctree)/kernel/configs/$@ $(srctree)/arch/$(SRCARCH)/c
 	$(if $(call configfiles),, $(error No configuration exists for this target on this architecture))
 	$(Q)$(CONFIG_SHELL) $(srctree)/scripts/kconfig/merge_config.sh -m .config $(configfiles)
 	+$(Q)yes "" | $(MAKE) -f $(srctree)/Makefile oldconfig
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 PHONY += kvmconfig
 kvmconfig: kvm_guest.config

@@ -130,6 +139,7 @@ xenconfig: xen.config
 PHONY += tinyconfig
 tinyconfig:
 	$(Q)$(MAKE) -f $(srctree)/Makefile allnoconfig tiny.config
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 # Help text used by make help
 help:

[email protected]

2017-12-29 18:32 UTC #5 lore raw

Quoted 7 lines · Paul Gortmaker

Quoted 5 lines · [Re: [Regression 4.15] Can't kill CONFIG_UNWINDER_ORC with fire or plague.] On…
This seems to be related to a kconfig quirk where only silentoldconfig updates the include/config/auto.conf file. The other config targets (oldconfig, defconfig, etc) don't touch it. It seems intentional, but I have no idea why.

Quoted 7 lines

Quoted 3 lines
That causes the Makefile to get stale data for 'CONFIG_*' variables when it includes auto.conf. So I don't think this is specific to the ORC check. It seems like it could also cause bugs elsewhere.
OK, good - you agree with my initial diagnosis of stale auto.conf then. Not sure what Randy was testing when he said he couldn't reproduce it.

FWIW just to add another data point, I tripped over this repeatedly during a recent bisect of v4.14..v4.15-rc4.

My ad-hoc solution was to just delete the stale auto.conf and reconfigure, IIRC. It was pretty annoying at the time, and I assumed everyone would be tripping over it so didn't bother making noise on lkml. Looks like I was mistaken...

Cheers,
Vito Caputo

Quoted 76 lines

Quoted 2 lines

The below (ugly) patch fixes it, though I'm not sure this is the best way to do it. We probably need Masahiro or Michal to chime in here.

Paul.
--

Quoted 65 lines

scripts/kconfig/Makefile +10 -0

index 297c1bf35140..3a0d92040a9c 100644
@@ -21,18 +21,23 @@ unexport CONFIG_
 
 xconfig: $(obj)/qconf
 	$< $(silent) $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 gconfig: $(obj)/gconf
 	$< $(silent) $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 menuconfig: $(obj)/mconf
 	$< $(silent) $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 config: $(obj)/conf
 	$< $(silent) --oldaskconfig $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 nconfig: $(obj)/nconf
 	$< $(silent) $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 silentoldconfig: $(obj)/conf
 	$(Q)mkdir -p include/config include/generated

@@ -85,6 +90,7 @@ PHONY += $(simple-targets)
 
 $(simple-targets): $(obj)/conf
 	$< $(silent) --$@ $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 PHONY += oldnoconfig savedefconfig defconfig

@@ -108,9 +114,11 @@ else
 	$(Q)$(MAKE) -f $(srctree)/Makefile $(KBUILD_DEFCONFIG)
 endif
 endif
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 %_defconfig: $(obj)/conf
 	$(Q)$< $(silent) --defconfig=arch/$(SRCARCH)/configs/$@ $(Kconfig)
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 configfiles=$(wildcard $(srctree)/kernel/configs/$@ $(srctree)/arch/$(SRCARCH)/configs/$@)

@@ -118,6 +126,7 @@ configfiles=$(wildcard $(srctree)/kernel/configs/$@ $(srctree)/arch/$(SRCARCH)/c
 	$(if $(call configfiles),, $(error No configuration exists for this target on this architecture))
 	$(Q)$(CONFIG_SHELL) $(srctree)/scripts/kconfig/merge_config.sh -m .config $(configfiles)
 	+$(Q)yes "" | $(MAKE) -f $(srctree)/Makefile oldconfig
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 PHONY += kvmconfig
 kvmconfig: kvm_guest.config

@@ -130,6 +139,7 @@ xenconfig: xen.config
 PHONY += tinyconfig
 tinyconfig:
 	$(Q)$(MAKE) -f $(srctree)/Makefile allnoconfig tiny.config
+	$(Q)$(MAKE) -f $(srctree)/Makefile silentoldconfig
 
 # Help text used by make help
 help:

Paul Gortmaker

author

2018-12-16 21:44 UTC #6 lore raw

In reply to Paul Gortmaker

Quoted 14 lines · [Re: [Regression 4.15] Can't kill CONFIG_UNWINDER_ORC with fire or plague.] On…

Quoted 8 lines · [Re: [Regression 4.15] Can't kill CONFIG_UNWINDER_ORC with fire or plague.] On…
This seems to be related to a kconfig quirk where only silentoldconfig updates the include/config/auto.conf file. The other config targets (oldconfig, defconfig, etc) don't touch it. It seems intentional, but I have no idea why.
That causes the Makefile to get stale data for 'CONFIG_*' variables when it includes auto.conf. So I don't think this is specific to the ORC check. It seems like it could also cause bugs elsewhere.
OK, good - you agree with my initial diagnosis of stale auto.conf then. Not sure what Randy was testing when he said he couldn't reproduce it.

Quoted 7 lines

Quoted 2 lines
The below (ugly) patch fixes it, though I'm not sure this is the best way to do it. We probably need Masahiro or Michal to chime in here.
Yep, that is why I intentionally put the kbuild folks on the To/Cc of the original report (and ran away screaming at the prospect of debugging Makefiles on xmas day). But with holidays and all, it might not be until early January before they have a chance to reply.

It is nearly a year later and we still have this false positive.

paul@sm:~/git/linux-head$ make -j12 > /dev/null
Makefile:966: *** "Cannot generate ORC metadata for CONFIG_UNWINDER_ORC=y,
 please install libelf-dev, libelf-devel or elfutils-libelf-devel".  Stop.
paul@sm:~/git/linux-head$ grep UNWINDER_ORC .config
# CONFIG_UNWINDER_ORC is not set

We do know a bit more now -- the auto.conf issue has been independently confirmed and "fixed" for other subsystems/issues since, like RETPOLINE:

---------------------
commit 25896d073d8a0403b07e6dec56f58e6c33678207
Author: Masahiro Yamada <[email protected]>
Date: Wed Dec 5 15:27:19 2018 +0900

x86/build: Fix compiler support check for CONFIG_RETPOLINE

It is troublesome to add a diagnostic like this to the Makefile parse stage because the top-level Makefile could be parsed with a stale include/config/auto.conf.

Once you are hit by the error about non-retpoline compiler, the compilation still breaks even after disabling CONFIG_RETPOLINE.
---------------------

I'm not sure if we want to treat this on a per config option each time again and again, or undertake a more global kbuild approach, but it does warrant a mention and a re-examination before we "solve" this again.

Paul.
--

Masahiro Yamada

2018-12-18 04:57 UTC #7 lore raw

Quoted 47 lines · Paul Gortmaker

Quoted 14 lines · [Re: [Regression 4.15] Can't kill CONFIG_UNWINDER_ORC with fire or plague.] On…
Quoted 8 lines · [Re: [Regression 4.15] Can't kill CONFIG_UNWINDER_ORC with fire or plague.] On…
This seems to be related to a kconfig quirk where only silentoldconfig updates the include/config/auto.conf file. The other config targets (oldconfig, defconfig, etc) don't touch it. It seems intentional, but I have no idea why.
That causes the Makefile to get stale data for 'CONFIG_*' variables when it includes auto.conf. So I don't think this is specific to the ORC check. It seems like it could also cause bugs elsewhere.
OK, good - you agree with my initial diagnosis of stale auto.conf then. Not sure what Randy was testing when he said he couldn't reproduce it.
Quoted 7 lines
Quoted 2 lines
The below (ugly) patch fixes it, though I'm not sure this is the best way to do it. We probably need Masahiro or Michal to chime in here.
Yep, that is why I intentionally put the kbuild folks on the To/Cc of the original report (and ran away screaming at the prospect of debugging Makefiles on xmas day). But with holidays and all, it might not be until early January before they have a chance to reply.
It is nearly a year later and we still have this false positive.
paul@sm:~/git/linux-head$ make -j12 > /dev/null
Makefile:966: *** "Cannot generate ORC metadata for CONFIG_UNWINDER_ORC=y,
 please install libelf-dev, libelf-devel or elfutils-libelf-devel".  Stop.
paul@sm:~/git/linux-head$ grep UNWINDER_ORC .config
# CONFIG_UNWINDER_ORC is not set
We do know a bit more now -- the auto.conf issue has been independently confirmed and "fixed" for other subsystems/issues since, like RETPOLINE:
---------------------
commit 25896d073d8a0403b07e6dec56f58e6c33678207
Author: Masahiro Yamada <[email protected]>
Date: Wed Dec 5 15:27:19 2018 +0900
x86/build: Fix compiler support check for CONFIG_RETPOLINE
It is troublesome to add a diagnostic like this to the Makefile parse stage because the top-level Makefile could be parsed with a stale include/config/auto.conf.

Quoted 7 lines

Once you are hit by the error about non-retpoline compiler, the compilation still breaks even after disabling CONFIG_RETPOLINE.
---------------------
I'm not sure if we want to treat this on a per config option each time again and again, or undertake a more global kbuild approach, but it does warrant a mention and a re-examination before we "solve" this again.

I did not notice this thread
(perhaps, it fell into my crack during the holidays) but I actually tried to fix this twice in a sophisticated way in the past.

The first attempt (https://patchwork.kernel.org/patch/10516049/)
was rejected by Josh Poimboeuf.

The second one (https://patchwork.kernel.org/patch/10643245/)
turned out not working as expected.

Now, I am preparing for the third attempt,
but it will take time for review.

What I can do now is the similar fix-up as commit 25896d073d8.

I will post a cheesy fix-up patch.

--
Best Regards
Masahiro Yamada