initramfs: Fix disabling of initramfs (and its compression)
PATCH 1/1 initramfs: Fix disabling of initramfs (and its compression)
Commit db2aa7fd15e8 ("initramfs: allow again choice of the embedded initram compression algorithm") introduced the possibility to select the initramfs compression algorithm from Kconfig and while this is a nice feature it broke the use case described below.
Here is what my build system does:
- kernel is initially configured not to have an initramfs included
- build the user space root file system
- re-configure the kernel to have an initramfs included (CONFIG_INITRAMFS_SOURCE="/path/to/romfs") and set relevant CONFIG_INITRAMFS options, in my case, no compression option (CONFIG_INITRAMFS_COMPRESSION_NONE)
- kernel is re-built with these options -> kernel+initramfs image is copied
- kernel is re-built again without these options -> kernel image is copied
Building a kernel without an initramfs means setting this option:
CONFIG_INITRAMFS_SOURCE="" (and this one only)
whereas building a kernel with an initramfs means setting these options:
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs /home/fainelli/work/uclinux-rootfs/misc/initramfs.dev" CONFIG_INITRAMFS_ROOT_UID=1000 CONFIG_INITRAMFS_ROOT_GID=1000 CONFIG_INITRAMFS_COMPRESSION_NONE=y CONFIG_INITRAMFS_COMPRESSION=""
Commit db2aa7fd15e857891cefbada8348c8d938c7a2bc ("initramfs: allow again choice of the embedded initram compression algorithm") is problematic because CONFIG_INITRAMFS_COMPRESSION which is used to determine the initramfs_data.cpio extension/compression is a string, and due to how Kconfig works it will evaluate in order, how to assign it.
Setting CONFIG_INITRAMFS_COMPRESSION_NONE with CONFIG_INITRAMFS_SOURCE="" cannot possibly work (because of the depends on INITRAMFS_SOURCE!="" imposed on CONFIG_INITRAMFS_COMPRESSION ) yet we still get CONFIG_INITRAMFS_COMPRESSION assigned to ".gz" because CONFIG_RD_GZIP=y is set in my kernel, even when there is no initramfs being built.
So we basically end-up generating two initramfs_data.cpio* files, one without extension, and one with .gz. This causes usr/Makefile to track usr/initramfs_data.cpio.gz, and not usr/initramfs_data.cpio anymore, that is also largely problematic after 9e3596b0c6539e28546ff7c72a06576627068353 ("kbuild: initramfs cleanup, set target from Kconfig") because we used to track all possible initramfs_data files in the $(targets) variable before that commit.
The end result is that the kernel with an initramfs clearly does not contain what we expect it to, it has a stale initramfs_data.cpio file built into it, and we keep re-generating an initramfs_data.cpio.gz file which is not the one that we want to include in the kernel image proper.
The fix consists in hiding CONFIG_INITRAMFS_COMPRESSION when CONFIG_INITRAMFS_SOURCE="". This puts us back in a state to the pre-4.10 behavior where we can properly disable and re-enable initramfs within the same kernel .config file, and be in control of what CONFIG_INITRAMFS_COMPRESSION is set to.
--- usr/Kconfig | 1 + 1 file changed, 1 insertion(+)
usr/Kconfig +1 -0
@@ -220,6 +220,7 @@ config INITRAMFS_COMPRESSION_LZ4 endchoice config INITRAMFS_COMPRESSION + depends on INITRAMFS_SOURCE!="" string default "" if INITRAMFS_COMPRESSION_NONE default ".gz" if INITRAMFS_COMPRESSION_GZIP
Signature · 1 line
2.9.3
Quoted 63 lines · On Sat, 20 May 2017 20:33:35 -0700 Florian Fainelli <[email protected]>…
Commit
db2aa7fd15e8("initramfs: allow again choice of the embedded initram compression algorithm") introduced the possibility to select the initramfs compression algorithm fromKconfigand while this is a nice feature it broke the use case described below.Here is what my build system does:
- kernel is initially configured not to have an initramfs included
- build the user space root file system
- re-configure the kernel to have an initramfs included (CONFIG_INITRAMFS_SOURCE="/path/to/romfs") and set relevantCONFIG_INITRAMFSoptions, in my case, no compression option (CONFIG_INITRAMFS_COMPRESSION_NONE)
- kernel is re-built with these options -> kernel+initramfs image is copied
- kernel is re-built again without these options -> kernel image is copiedBuilding a kernel without an initramfs means setting this option:
CONFIG_INITRAMFS_SOURCE="" (and this one only)whereas building a kernel with an initramfs means setting these options:
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000CONFIG_INITRAMFS_COMPRESSION_NONE=yCONFIG_INITRAMFS_COMPRESSION=""Commit
db2aa7fd15e857891cefbada8348c8d938c7a2bc("initramfs: allow again choice of the embedded initram compression algorithm") is problematic becauseCONFIG_INITRAMFS_COMPRESSIONwhich is used to determine the initramfs_data.cpio extension/compression is a string, and due to howKconfigworks it will evaluate in order, how to assign it.Setting
CONFIG_INITRAMFS_COMPRESSION_NONEwithCONFIG_INITRAMFS_SOURCE="" cannot possibly work (because of the depends on INITRAMFS_SOURCE!="" imposed onCONFIG_INITRAMFS_COMPRESSION) yet we still getCONFIG_INITRAMFS_COMPRESSIONassigned to ".gz" becauseCONFIG_RD_GZIP=yis set in my kernel, even when there is no initramfs being built.So we basically end-up generating two initramfs_data.cpio* files, one without extension, and one with .gz. This causes usr/Makefile to track
usr/initramfs_data.cpio.gz, and notusr/initramfs_data.cpioanymore, that is also largely problematic after9e3596b0c6539e28546ff7c72a06576627068353("kbuild: initramfs cleanup, set target fromKconfig") because we used to track all possibleinitramfs_datafiles in the $(targets) variable before that commit.The end result is that the kernel with an initramfs clearly does not contain what we expect it to, it has a stale initramfs_data.cpio file built into it, and we keep re-generating an initramfs_data.cpio.gz file which is not the one that we want to include in the kernel image proper.
The fix consists in hiding
CONFIG_INITRAMFS_COMPRESSIONwhenCONFIG_INITRAMFS_SOURCE="". This puts us back in a state to the pre-4.10 behavior where we can properly disable and re-enable initramfs within the same kernel .config file, and be in control of whatCONFIG_INITRAMFS_COMPRESSIONis set to.
Quoted 3 lines
Fixes db2aa7fd15e8 ("initramfs: allow again choice of the embedded initram compression algorithm")Fixes 9e3596b0c653 ("kbuild: initramfs cleanup, set target from Kconfig")Signed-off-by Florian Fainelli <[email protected]>
This is very thorough, thank you for tracking it down and fixing it.
I can't say I've worked through the problem in the code, but your changelog and the proposed fix seem reasonable to me. So for what it's worth:
Quoted 16 lines
--- usr/Kconfig | 1 + 1 file changed, 1 insertion(+)usr/Kconfig +1
-0@@ -220,6 +220,7 @@ config INITRAMFS_COMPRESSION_LZ4 endchoice config INITRAMFS_COMPRESSION + depends on INITRAMFS_SOURCE!="" string default "" if INITRAMFS_COMPRESSION_NONE default ".gz" if INITRAMFS_COMPRESSION_GZIP
Quoted 70 lines · Nicholas Piggin
Quoted 63 lines · On Sat, 20 May 2017 20:33:35 -0700 Florian Fainelli <[email protected]>…
Commit
db2aa7fd15e8("initramfs: allow again choice of the embedded initram compression algorithm") introduced the possibility to select the initramfs compression algorithm fromKconfigand while this is a nice feature it broke the use case described below.Here is what my build system does:
- kernel is initially configured not to have an initramfs included
- build the user space root file system
- re-configure the kernel to have an initramfs included (CONFIG_INITRAMFS_SOURCE="/path/to/romfs") and set relevantCONFIG_INITRAMFSoptions, in my case, no compression option (CONFIG_INITRAMFS_COMPRESSION_NONE)
- kernel is re-built with these options -> kernel+initramfs image is copied
- kernel is re-built again without these options -> kernel image is copiedBuilding a kernel without an initramfs means setting this option:
CONFIG_INITRAMFS_SOURCE="" (and this one only)whereas building a kernel with an initramfs means setting these options:
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000CONFIG_INITRAMFS_COMPRESSION_NONE=yCONFIG_INITRAMFS_COMPRESSION=""Commit
db2aa7fd15e857891cefbada8348c8d938c7a2bc("initramfs: allow again choice of the embedded initram compression algorithm") is problematic becauseCONFIG_INITRAMFS_COMPRESSIONwhich is used to determine the initramfs_data.cpio extension/compression is a string, and due to howKconfigworks it will evaluate in order, how to assign it.Setting
CONFIG_INITRAMFS_COMPRESSION_NONEwithCONFIG_INITRAMFS_SOURCE="" cannot possibly work (because of the depends on INITRAMFS_SOURCE!="" imposed onCONFIG_INITRAMFS_COMPRESSION) yet we still getCONFIG_INITRAMFS_COMPRESSIONassigned to ".gz" becauseCONFIG_RD_GZIP=yis set in my kernel, even when there is no initramfs being built.So we basically end-up generating two initramfs_data.cpio* files, one without extension, and one with .gz. This causes usr/Makefile to track
usr/initramfs_data.cpio.gz, and notusr/initramfs_data.cpioanymore, that is also largely problematic after9e3596b0c6539e28546ff7c72a06576627068353("kbuild: initramfs cleanup, set target fromKconfig") because we used to track all possibleinitramfs_datafiles in the $(targets) variable before that commit.The end result is that the kernel with an initramfs clearly does not contain what we expect it to, it has a stale initramfs_data.cpio file built into it, and we keep re-generating an initramfs_data.cpio.gz file which is not the one that we want to include in the kernel image proper.
The fix consists in hiding
CONFIG_INITRAMFS_COMPRESSIONwhenCONFIG_INITRAMFS_SOURCE="". This puts us back in a state to the pre-4.10 behavior where we can properly disable and re-enable initramfs within the same kernel .config file, and be in control of whatCONFIG_INITRAMFS_COMPRESSIONis set to.Quoted 3 lines
Fixes db2aa7fd15e8 ("initramfs: allow again choice of the embedded initram compression algorithm")Fixes 9e3596b0c653 ("kbuild: initramfs cleanup, set target from Kconfig")Signed-off-by Florian Fainelli <[email protected]>
Quoted 7 lines
This is very thorough, thank you for tracking it down and fixing it.
I can't say I've worked through the problem in the code, but your changelog and the proposed fix seem reasonable to me. So for what it's worth:
Acked-by Nicholas Piggin <[email protected]>
Well, I am looking at this again, and it's still broken, the same test case is involved, except this time, I am switching beween no-initramfs and initramfs with gzip compression (the key thing is using a compression of some sort). The end result is the following:
- change stuff in the rootfs
- build the kernel with initramfs, CONFIG_INITRAMFS_COMPRESSION_GZIP=y, usr/initramfs_data.cpio.gz gets generated correctly the first time
- build the kernel without initramfs, CONFIG_INITRAMFS_COMPRESSION_NONE=y, usr/initramfs_data.cpio gets generated
Now back to step 1 add some files, and we can see that usr/initramfs_data.cpio.gz is now stale from before...
So while my earlier fix switched the initramfs w/o compression to no initramfs rebuild, now this does not work because we still have two files left to be tracked:
usr/initramfs_data.cpio (no compression, or when initramfs is disabled) and usr/initramfs_data.cpio.$(suffix_y)
How would you go about solving this?
Quoted 18 lines
Quoted 17 lines
--- usr/Kconfig | 1 + 1 file changed, 1 insertion(+)usr/Kconfig +1
-0@@ -220,6 +220,7 @@ config INITRAMFS_COMPRESSION_LZ4 endchoice config INITRAMFS_COMPRESSION + depends on INITRAMFS_SOURCE!="" string default "" if INITRAMFS_COMPRESSION_NONE default ".gz" if INITRAMFS_COMPRESSION_GZIP
Signature · 1 line
Florian
Quoted 93 lines · On Mon, 28 Aug 2017 13:03:31 -0700 Florian Fainelli <[email protected]>…
Quoted 70 lines · Nicholas Piggin
Quoted 63 lines · On Sat, 20 May 2017 20:33:35 -0700 Florian Fainelli <[email protected]>…
Commit
db2aa7fd15e8("initramfs: allow again choice of the embedded initram compression algorithm") introduced the possibility to select the initramfs compression algorithm fromKconfigand while this is a nice feature it broke the use case described below.Here is what my build system does:
- kernel is initially configured not to have an initramfs included
- build the user space root file system
- re-configure the kernel to have an initramfs included (CONFIG_INITRAMFS_SOURCE="/path/to/romfs") and set relevantCONFIG_INITRAMFSoptions, in my case, no compression option (CONFIG_INITRAMFS_COMPRESSION_NONE)
- kernel is re-built with these options -> kernel+initramfs image is copied
- kernel is re-built again without these options -> kernel image is copiedBuilding a kernel without an initramfs means setting this option:
CONFIG_INITRAMFS_SOURCE="" (and this one only)whereas building a kernel with an initramfs means setting these options:
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000CONFIG_INITRAMFS_COMPRESSION_NONE=yCONFIG_INITRAMFS_COMPRESSION=""Commit
db2aa7fd15e857891cefbada8348c8d938c7a2bc("initramfs: allow again choice of the embedded initram compression algorithm") is problematic becauseCONFIG_INITRAMFS_COMPRESSIONwhich is used to determine the initramfs_data.cpio extension/compression is a string, and due to howKconfigworks it will evaluate in order, how to assign it.Setting
CONFIG_INITRAMFS_COMPRESSION_NONEwithCONFIG_INITRAMFS_SOURCE="" cannot possibly work (because of the depends on INITRAMFS_SOURCE!="" imposed onCONFIG_INITRAMFS_COMPRESSION) yet we still getCONFIG_INITRAMFS_COMPRESSIONassigned to ".gz" becauseCONFIG_RD_GZIP=yis set in my kernel, even when there is no initramfs being built.So we basically end-up generating two initramfs_data.cpio* files, one without extension, and one with .gz. This causes usr/Makefile to track
usr/initramfs_data.cpio.gz, and notusr/initramfs_data.cpioanymore, that is also largely problematic after9e3596b0c6539e28546ff7c72a06576627068353("kbuild: initramfs cleanup, set target fromKconfig") because we used to track all possibleinitramfs_datafiles in the $(targets) variable before that commit.The end result is that the kernel with an initramfs clearly does not contain what we expect it to, it has a stale initramfs_data.cpio file built into it, and we keep re-generating an initramfs_data.cpio.gz file which is not the one that we want to include in the kernel image proper.
The fix consists in hiding
CONFIG_INITRAMFS_COMPRESSIONwhenCONFIG_INITRAMFS_SOURCE="". This puts us back in a state to the pre-4.10 behavior where we can properly disable and re-enable initramfs within the same kernel .config file, and be in control of whatCONFIG_INITRAMFS_COMPRESSIONis set to.Quoted 3 lines
Fixes db2aa7fd15e8 ("initramfs: allow again choice of the embedded initram compression algorithm")Fixes 9e3596b0c653 ("kbuild: initramfs cleanup, set target from Kconfig")Signed-off-by Florian Fainelli <[email protected]>Quoted 7 lines
This is very thorough, thank you for tracking it down and fixing it.
I can't say I've worked through the problem in the code, but your changelog and the proposed fix seem reasonable to me. So for what it's worth:
Acked-by Nicholas Piggin <[email protected]>Well, I am looking at this again, and it's still broken, the same test case is involved, except this time, I am switching beween no-initramfs and initramfs with gzip compression (the key thing is using a compression of some sort). The end result is the following:
- change stuff in the rootfs
- build the kernel with initramfs,CONFIG_INITRAMFS_COMPRESSION_GZIP=y,usr/initramfs_data.cpio.gzgets generated correctly the first time
- build the kernel without initramfs,CONFIG_INITRAMFS_COMPRESSION_NONE=y,usr/initramfs_data.cpiogets generatedNow back to step 1 add some files, and we can see that
usr/initramfs_data.cpio.gzis now stale from before...
Quoted 8 lines
So while my earlier fix switched the initramfs w/o compression to no initramfs rebuild, now this does not work because we still have two files left to be tracked:
usr/initramfs_data.cpio(no compression, or when initramfs is disabled) and usr/initramfs_data.cpio.$(suffix_y)How would you go about solving this?
I don't see the problem. When I change back to .gz, modify the source directory, and rebuild, it rebuilds a new initramfs_data.cpio.gz here.
Thanks,
Nick
Quoted 100 lines · Nicholas Piggin
Quoted 97 lines · On Mon, 28 Aug 2017 13:03:31 -0700 Florian Fainelli <[email protected]>…
Quoted 70 lines · Nicholas Piggin
Quoted 63 lines · On Sat, 20 May 2017 20:33:35 -0700 Florian Fainelli <[email protected]>…
Commit
db2aa7fd15e8("initramfs: allow again choice of the embedded initram compression algorithm") introduced the possibility to select the initramfs compression algorithm fromKconfigand while this is a nice feature it broke the use case described below.Here is what my build system does:
- kernel is initially configured not to have an initramfs included
- build the user space root file system
- re-configure the kernel to have an initramfs included (CONFIG_INITRAMFS_SOURCE="/path/to/romfs") and set relevantCONFIG_INITRAMFSoptions, in my case, no compression option (CONFIG_INITRAMFS_COMPRESSION_NONE)
- kernel is re-built with these options -> kernel+initramfs image is copied
- kernel is re-built again without these options -> kernel image is copiedBuilding a kernel without an initramfs means setting this option:
CONFIG_INITRAMFS_SOURCE="" (and this one only)whereas building a kernel with an initramfs means setting these options:
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000CONFIG_INITRAMFS_COMPRESSION_NONE=yCONFIG_INITRAMFS_COMPRESSION=""Commit
db2aa7fd15e857891cefbada8348c8d938c7a2bc("initramfs: allow again choice of the embedded initram compression algorithm") is problematic becauseCONFIG_INITRAMFS_COMPRESSIONwhich is used to determine the initramfs_data.cpio extension/compression is a string, and due to howKconfigworks it will evaluate in order, how to assign it.Setting
CONFIG_INITRAMFS_COMPRESSION_NONEwithCONFIG_INITRAMFS_SOURCE="" cannot possibly work (because of the depends on INITRAMFS_SOURCE!="" imposed onCONFIG_INITRAMFS_COMPRESSION) yet we still getCONFIG_INITRAMFS_COMPRESSIONassigned to ".gz" becauseCONFIG_RD_GZIP=yis set in my kernel, even when there is no initramfs being built.So we basically end-up generating two initramfs_data.cpio* files, one without extension, and one with .gz. This causes usr/Makefile to track
usr/initramfs_data.cpio.gz, and notusr/initramfs_data.cpioanymore, that is also largely problematic after9e3596b0c6539e28546ff7c72a06576627068353("kbuild: initramfs cleanup, set target fromKconfig") because we used to track all possibleinitramfs_datafiles in the $(targets) variable before that commit.The end result is that the kernel with an initramfs clearly does not contain what we expect it to, it has a stale initramfs_data.cpio file built into it, and we keep re-generating an initramfs_data.cpio.gz file which is not the one that we want to include in the kernel image proper.
The fix consists in hiding
CONFIG_INITRAMFS_COMPRESSIONwhenCONFIG_INITRAMFS_SOURCE="". This puts us back in a state to the pre-4.10 behavior where we can properly disable and re-enable initramfs within the same kernel .config file, and be in control of whatCONFIG_INITRAMFS_COMPRESSIONis set to.Quoted 3 lines
Fixes db2aa7fd15e8 ("initramfs: allow again choice of the embedded initram compression algorithm")Fixes 9e3596b0c653 ("kbuild: initramfs cleanup, set target from Kconfig")Signed-off-by Florian Fainelli <[email protected]>Quoted 7 lines
This is very thorough, thank you for tracking it down and fixing it.
I can't say I've worked through the problem in the code, but your changelog and the proposed fix seem reasonable to me. So for what it's worth:
Acked-by Nicholas Piggin <[email protected]>Well, I am looking at this again, and it's still broken, the same test case is involved, except this time, I am switching beween no-initramfs and initramfs with gzip compression (the key thing is using a compression of some sort). The end result is the following:
- change stuff in the rootfs
- build the kernel with initramfs,CONFIG_INITRAMFS_COMPRESSION_GZIP=y,usr/initramfs_data.cpio.gzgets generated correctly the first time
- build the kernel without initramfs,CONFIG_INITRAMFS_COMPRESSION_NONE=y,usr/initramfs_data.cpiogets generatedNow back to step 1 add some files, and we can see that
usr/initramfs_data.cpio.gzis now stale from before...So while my earlier fix switched the initramfs w/o compression to no initramfs rebuild, now this does not work because we still have two files left to be tracked:
Quoted 7 lines
Quoted 4 lines
usr/initramfs_data.cpio(no compression, or when initramfs is disabled) and usr/initramfs_data.cpio.$(suffix_y)How would you go about solving this?
I don't see the problem. When I change back to .gz, modify the source directory, and rebuild, it rebuilds a new initramfs_data.cpio.gz here.
Can you share the different steps involved here? Here is what the build system I am using is doing.
initramfs:
https://github.com/Broadcom/stblinux-4.9/blob/master/rootfs/Makefile#L169
no-initramfs:
https://github.com/Broadcom/stblinux-4.9/blob/master/rootfs/Makefile#L189
The script that does the config option switching is here:
https://github.com/Broadcom/stblinux-4.9/blob/master/rootfs/bin/config.pl#L1072
When you say change the source directory, you mean setting CONFIG_INITRAMFS_SOURCE to one directory that contains your rootfs, and then setting it to empty to disable that, right?
Here is what I have so far, only caveat (still triple checking) is that we seem to be always re-generating usr/initramfs_cpio.$(suffix_y) which is still an improvement over the current situation where we have a stale file after the first creation.
usr/Makefile +1 -0
@@ -7,6 +7,7 @@ PHONY += klibcdirs suffix_y = $(subst $\",,$(CONFIG_INITRAMFS_COMPRESSION)) datafile_y = initramfs_data.cpio$(suffix_y) +datafile_d_y = .$(datafile_y).d AFLAGS_initramfs_data.o += -DINITRAMFS_IMAGE="usr/$(datafile_y)"
@@ -29,12 +30,12 @@ ramfs-args := \ $(if $(CONFIG_INITRAMFS_ROOT_UID), -u
$(CONFIG_INITRAMFS_ROOT_UID)) \
$(if $(CONFIG_INITRAMFS_ROOT_GID), -g $(CONFIG_INITRAMFS_ROOT_GID))-# .initramfs_data.cpio.d is used to identify all files included +# $(datafile_d_y) is used to identify all files included
# in initramfs and to detect if any files are added/removed. # Removed files are identified by directory timestamp being updated # The dependency list is generated by gen_initramfs.sh -l -ifneq ($(wildcard $(obj)/.initramfs_data.cpio.d),) - include $(obj)/.initramfs_data.cpio.d +ifneq ($(wildcard $(obj)/$(datafile_d_y)),) + include $(obj)/$(datafile_d_y) endif
quiet_cmd_initfs = GEN $@
@@ -52,5 +53,5 @@ $(deps_initramfs): klibcdirs
# 3) If gen_init_cpio are newer than initramfs_data.cpio
# 4) arguments to gen_initramfs.sh changes $(obj)/$(datafile_y): $(obj)/gen_init_cpio $(deps_initramfs) klibcdirs
- $(Q)$(initramfs) -l $(ramfs-input) > $(obj)/.initramfs_data.cpio.d
+ $(Q)$(initramfs) -l $(ramfs-input) > $(obj)/$(datafile_d_y)
$(call if_changed,initfs)
Signature · 1 line
Florian
Quoted 100 lines · Nicholas Piggin
Quoted 97 lines · On Mon, 28 Aug 2017 13:03:31 -0700 Florian Fainelli <[email protected]>…
Quoted 70 lines · Nicholas Piggin
Quoted 63 lines · On Sat, 20 May 2017 20:33:35 -0700 Florian Fainelli <[email protected]>…
Commit
db2aa7fd15e8("initramfs: allow again choice of the embedded initram compression algorithm") introduced the possibility to select the initramfs compression algorithm fromKconfigand while this is a nice feature it broke the use case described below.Here is what my build system does:
- kernel is initially configured not to have an initramfs included
- build the user space root file system
- re-configure the kernel to have an initramfs included (CONFIG_INITRAMFS_SOURCE="/path/to/romfs") and set relevantCONFIG_INITRAMFSoptions, in my case, no compression option (CONFIG_INITRAMFS_COMPRESSION_NONE)
- kernel is re-built with these options -> kernel+initramfs image is copied
- kernel is re-built again without these options -> kernel image is copiedBuilding a kernel without an initramfs means setting this option:
CONFIG_INITRAMFS_SOURCE="" (and this one only)whereas building a kernel with an initramfs means setting these options:
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000CONFIG_INITRAMFS_COMPRESSION_NONE=yCONFIG_INITRAMFS_COMPRESSION=""Commit
db2aa7fd15e857891cefbada8348c8d938c7a2bc("initramfs: allow again choice of the embedded initram compression algorithm") is problematic becauseCONFIG_INITRAMFS_COMPRESSIONwhich is used to determine the initramfs_data.cpio extension/compression is a string, and due to howKconfigworks it will evaluate in order, how to assign it.Setting
CONFIG_INITRAMFS_COMPRESSION_NONEwithCONFIG_INITRAMFS_SOURCE="" cannot possibly work (because of the depends on INITRAMFS_SOURCE!="" imposed onCONFIG_INITRAMFS_COMPRESSION) yet we still getCONFIG_INITRAMFS_COMPRESSIONassigned to ".gz" becauseCONFIG_RD_GZIP=yis set in my kernel, even when there is no initramfs being built.So we basically end-up generating two initramfs_data.cpio* files, one without extension, and one with .gz. This causes usr/Makefile to track
usr/initramfs_data.cpio.gz, and notusr/initramfs_data.cpioanymore, that is also largely problematic after9e3596b0c6539e28546ff7c72a06576627068353("kbuild: initramfs cleanup, set target fromKconfig") because we used to track all possibleinitramfs_datafiles in the $(targets) variable before that commit.The end result is that the kernel with an initramfs clearly does not contain what we expect it to, it has a stale initramfs_data.cpio file built into it, and we keep re-generating an initramfs_data.cpio.gz file which is not the one that we want to include in the kernel image proper.
The fix consists in hiding
CONFIG_INITRAMFS_COMPRESSIONwhenCONFIG_INITRAMFS_SOURCE="". This puts us back in a state to the pre-4.10 behavior where we can properly disable and re-enable initramfs within the same kernel .config file, and be in control of whatCONFIG_INITRAMFS_COMPRESSIONis set to.Quoted 3 lines
Fixes db2aa7fd15e8 ("initramfs: allow again choice of the embedded initram compression algorithm")Fixes 9e3596b0c653 ("kbuild: initramfs cleanup, set target from Kconfig")Signed-off-by Florian Fainelli <[email protected]>Quoted 7 lines
This is very thorough, thank you for tracking it down and fixing it.
I can't say I've worked through the problem in the code, but your changelog and the proposed fix seem reasonable to me. So for what it's worth:
Acked-by Nicholas Piggin <[email protected]>Well, I am looking at this again, and it's still broken, the same test case is involved, except this time, I am switching beween no-initramfs and initramfs with gzip compression (the key thing is using a compression of some sort). The end result is the following:
- change stuff in the rootfs
- build the kernel with initramfs,CONFIG_INITRAMFS_COMPRESSION_GZIP=y,usr/initramfs_data.cpio.gzgets generated correctly the first time
- build the kernel without initramfs,CONFIG_INITRAMFS_COMPRESSION_NONE=y,usr/initramfs_data.cpiogets generatedNow back to step 1 add some files, and we can see that
usr/initramfs_data.cpio.gzis now stale from before...So while my earlier fix switched the initramfs w/o compression to no initramfs rebuild, now this does not work because we still have two files left to be tracked:
Quoted 7 lines
Quoted 4 lines
usr/initramfs_data.cpio(no compression, or when initramfs is disabled) and usr/initramfs_data.cpio.$(suffix_y)How would you go about solving this?
I don't see the problem. When I change back to .gz, modify the source directory, and rebuild, it rebuilds a new initramfs_data.cpio.gz here.
This is really puzzling me, so here is what is going on my side:
First time build w/ initramfs enabled:
grep INITRAMFS .config CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs /home/fainelli/work/uclinux-rootfs/misc/initramfs.dev" CONFIG_INITRAMFS_ROOT_UID=1000 CONFIG_INITRAMFS_ROOT_GID=1000
# CONFIG_INITRAMFS_COMPRESSION_NONE is not set CONFIG_INITRAMFS_COMPRESSION_GZIP=y
# CONFIG_INITRAMFS_COMPRESSION_BZIP2 is not set
# CONFIG_INITRAMFS_COMPRESSION_LZMA is not set
# CONFIG_INITRAMFS_COMPRESSION_XZ is not set
# CONFIG_INITRAMFS_COMPRESSION_LZO is not set
# CONFIG_INITRAMFS_COMPRESSION_LZ4 is not set CONFIG_INITRAMFS_COMPRESSION=".gz"
fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 6.8M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:37 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:38 ../ -rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:37 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:37 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 9.6K Sep 7 05:36 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37 .initramfs_data.cpio.gz.cmd -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:37 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:34 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:35 modules.order
Now let's reconfigure this kernel w/o initramfs and rebuild:
grep INITRAMFS .configCONFIG_INITRAMFS_SOURCE=""
fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 3.6M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:44 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:44 ../ -rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:44 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:44 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 512 Sep 7 05:44 initramfs_data.cpio -rw-rw-r-- 1 fainelli fainelli 105 Sep 7 05:44 .initramfs_data.cpio.cmd -rw-rw-r-- 1 fainelli fainelli 52 Sep 7 05:44 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37 .initramfs_data.cpio.gz.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Sep 7 05:44 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:44 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:34 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:35 modules.order
Now let's rebuild with a new application added to the root filesystem:
grep INITRAMFS .config CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs /home/fainelli/work/uclinux-rootfs/misc/initramfs.dev" CONFIG_INITRAMFS_ROOT_UID=1000 CONFIG_INITRAMFS_ROOT_GID=1000
# CONFIG_INITRAMFS_COMPRESSION_NONE is not set CONFIG_INITRAMFS_COMPRESSION_GZIP=y
# CONFIG_INITRAMFS_COMPRESSION_BZIP2 is not set
# CONFIG_INITRAMFS_COMPRESSION_LZMA is not set
# CONFIG_INITRAMFS_COMPRESSION_XZ is not set
# CONFIG_INITRAMFS_COMPRESSION_LZO is not set
# CONFIG_INITRAMFS_COMPRESSION_LZ4 is not set CONFIG_INITRAMFS_COMPRESSION=".gz"
We are not fine here:scripts/kconfig/conf --silentoldconfig Kconfig
#
# configuration written to .config # CHK include/config/kernel.release CHK include/generated/uapi/linux/version.h 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 AS usr/initramfs_data.o
^===== we should have called gzip to re-generate the initramfs, we did not
AR usr/built-in.o GEN .version CHK include/generated/compile.h UPD include/generated/compile.h CC init/version.o AR init/built-in.o AR built-in.o LD vmlinux.o
MODPOST vmlinux.o
And the timestamps here confirm that:
fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 6.9M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:46 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:46 ../ -rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:46 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:46 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 512 Sep 7 05:44 initramfs_data.cpio -rw-rw-r-- 1 fainelli fainelli 105 Sep 7 05:44 .initramfs_data.cpio.cmd -rw-rw-r-- 1 fainelli fainelli 11K Sep 7 05:46 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37 .initramfs_data.cpio.gz.cmd -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:46 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:46 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:46 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:46 modules.order
Worse yet, usr/.initramfs_data.cpio.d does list the binary that was just added, but for some reason, this did not trigger a rebuild of initramfs_data.cpio.gz do you see any reasons why?
Thanks!
Signature · 1 line
Florian
Quoted 248 lines · On Thu, 7 Sep 2017 05:50:30 -0700 Florian Fainelli <[email protected]> wrote:
Quoted 100 lines · Nicholas Piggin
Quoted 97 lines · On Mon, 28 Aug 2017 13:03:31 -0700 Florian Fainelli <[email protected]>…
Quoted 70 lines · Nicholas Piggin
Quoted 63 lines · On Sat, 20 May 2017 20:33:35 -0700 Florian Fainelli <[email protected]>…
Commit
db2aa7fd15e8("initramfs: allow again choice of the embedded initram compression algorithm") introduced the possibility to select the initramfs compression algorithm fromKconfigand while this is a nice feature it broke the use case described below.Here is what my build system does:
- kernel is initially configured not to have an initramfs included
- build the user space root file system
- re-configure the kernel to have an initramfs included (CONFIG_INITRAMFS_SOURCE="/path/to/romfs") and set relevantCONFIG_INITRAMFSoptions, in my case, no compression option (CONFIG_INITRAMFS_COMPRESSION_NONE)
- kernel is re-built with these options -> kernel+initramfs image is copied
- kernel is re-built again without these options -> kernel image is copiedBuilding a kernel without an initramfs means setting this option:
CONFIG_INITRAMFS_SOURCE="" (and this one only)whereas building a kernel with an initramfs means setting these options:
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000CONFIG_INITRAMFS_COMPRESSION_NONE=yCONFIG_INITRAMFS_COMPRESSION=""Commit
db2aa7fd15e857891cefbada8348c8d938c7a2bc("initramfs: allow again choice of the embedded initram compression algorithm") is problematic becauseCONFIG_INITRAMFS_COMPRESSIONwhich is used to determine the initramfs_data.cpio extension/compression is a string, and due to howKconfigworks it will evaluate in order, how to assign it.Setting
CONFIG_INITRAMFS_COMPRESSION_NONEwithCONFIG_INITRAMFS_SOURCE="" cannot possibly work (because of the depends on INITRAMFS_SOURCE!="" imposed onCONFIG_INITRAMFS_COMPRESSION) yet we still getCONFIG_INITRAMFS_COMPRESSIONassigned to ".gz" becauseCONFIG_RD_GZIP=yis set in my kernel, even when there is no initramfs being built.So we basically end-up generating two initramfs_data.cpio* files, one without extension, and one with .gz. This causes usr/Makefile to track
usr/initramfs_data.cpio.gz, and notusr/initramfs_data.cpioanymore, that is also largely problematic after9e3596b0c6539e28546ff7c72a06576627068353("kbuild: initramfs cleanup, set target fromKconfig") because we used to track all possibleinitramfs_datafiles in the $(targets) variable before that commit.The end result is that the kernel with an initramfs clearly does not contain what we expect it to, it has a stale initramfs_data.cpio file built into it, and we keep re-generating an initramfs_data.cpio.gz file which is not the one that we want to include in the kernel image proper.
The fix consists in hiding
CONFIG_INITRAMFS_COMPRESSIONwhenCONFIG_INITRAMFS_SOURCE="". This puts us back in a state to the pre-4.10 behavior where we can properly disable and re-enable initramfs within the same kernel .config file, and be in control of whatCONFIG_INITRAMFS_COMPRESSIONis set to.Quoted 3 lines
Fixes db2aa7fd15e8 ("initramfs: allow again choice of the embedded initram compression algorithm")Fixes 9e3596b0c653 ("kbuild: initramfs cleanup, set target from Kconfig")Signed-off-by Florian Fainelli <[email protected]>Quoted 7 lines
This is very thorough, thank you for tracking it down and fixing it.
I can't say I've worked through the problem in the code, but your changelog and the proposed fix seem reasonable to me. So for what it's worth:
Acked-by Nicholas Piggin <[email protected]>Well, I am looking at this again, and it's still broken, the same test case is involved, except this time, I am switching beween no-initramfs and initramfs with gzip compression (the key thing is using a compression of some sort). The end result is the following:
- change stuff in the rootfs
- build the kernel with initramfs,CONFIG_INITRAMFS_COMPRESSION_GZIP=y,usr/initramfs_data.cpio.gzgets generated correctly the first time
- build the kernel without initramfs,CONFIG_INITRAMFS_COMPRESSION_NONE=y,usr/initramfs_data.cpiogets generatedNow back to step 1 add some files, and we can see that
usr/initramfs_data.cpio.gzis now stale from before...So while my earlier fix switched the initramfs w/o compression to no initramfs rebuild, now this does not work because we still have two files left to be tracked:
Quoted 7 lines
Quoted 4 lines
usr/initramfs_data.cpio(no compression, or when initramfs is disabled) and usr/initramfs_data.cpio.$(suffix_y)How would you go about solving this?
I don't see the problem. When I change back to .gz, modify the source directory, and rebuild, it rebuilds a new initramfs_data.cpio.gz here.
This is really puzzling me, so here is what is going on my side:
First time build w/ initramfs enabled:
grep INITRAMFS .config
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000
#CONFIG_INITRAMFS_COMPRESSION_NONEis not setCONFIG_INITRAMFS_COMPRESSION_GZIP=y
#CONFIG_INITRAMFS_COMPRESSION_BZIP2is not set
#CONFIG_INITRAMFS_COMPRESSION_LZMAis not set
#CONFIG_INITRAMFS_COMPRESSION_XZis not set
#CONFIG_INITRAMFS_COMPRESSION_LZOis not set
#CONFIG_INITRAMFS_COMPRESSION_LZ4is not setCONFIG_INITRAMFS_COMPRESSION=".gz"fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 6.8M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:37 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:38 ../(quoted diff) +0
-9-rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:37 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:37 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 9.6K Sep 7 05:36 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37.initramfs_data.cpio.gz.cmd
(quoted diff) +0
-7-rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:37 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:34 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:35 modules.orderNow let's reconfigure this kernel w/o initramfs and rebuild:
grep INITRAMFS .config
CONFIG_INITRAMFS_SOURCE=""fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 3.6M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:44 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:44 ../(quoted diff) +0
-11-rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:44 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:44 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 512 Sep 7 05:44 initramfs_data.cpio -rw-rw-r-- 1 fainelli fainelli 105 Sep 7 05:44 .initramfs_data.cpio.cmd -rw-rw-r-- 1 fainelli fainelli 52 Sep 7 05:44 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37.initramfs_data.cpio.gz.cmd
(quoted diff) +0
-7-rw-rw-r-- 1 fainelli fainelli 1.3K Sep 7 05:44 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:44 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:34 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:35 modules.orderNow let's rebuild with a new application added to the root filesystem:
grep INITRAMFS .config
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000
#CONFIG_INITRAMFS_COMPRESSION_NONEis not setCONFIG_INITRAMFS_COMPRESSION_GZIP=y
#CONFIG_INITRAMFS_COMPRESSION_BZIP2is not set
#CONFIG_INITRAMFS_COMPRESSION_LZMAis not set
#CONFIG_INITRAMFS_COMPRESSION_XZis not set
#CONFIG_INITRAMFS_COMPRESSION_LZOis not set
#CONFIG_INITRAMFS_COMPRESSION_LZ4is not setCONFIG_INITRAMFS_COMPRESSION=".gz"We are not fine here:
scripts/kconfig/conf--silentoldconfigKconfig
## configuration written to .config # CHK include/config/kernel.release CHK include/generated/uapi/linux/version.h 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 AS usr/initramfs_data.o^===== we should have called gzip to re-generate the initramfs, we did not
AR usr/built-in.o GEN .version CHK include/generated/compile.h UPD include/generated/compile.h CC init/version.o AR init/built-in.o AR built-in.o LD vmlinux.oMODPOST vmlinux.o
And the timestamps here confirm that:
fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 6.9M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:46 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:46 ../(quoted diff) +0
-11-rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:46 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:46 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 512 Sep 7 05:44 initramfs_data.cpio -rw-rw-r-- 1 fainelli fainelli 105 Sep 7 05:44 .initramfs_data.cpio.cmd -rw-rw-r-- 1 fainelli fainelli 11K Sep 7 05:46 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37.initramfs_data.cpio.gz.cmd
(quoted diff) +0
-7-rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:46 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:46 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:46 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:46 modules.order
Quoted 3 lines
Worse yet,
usr/.initramfs_data.cpio.ddoes list the binary that was just added, but for some reason, this did not trigger a rebuild of initramfs_data.cpio.gz do you see any reasons why?
This rule
$(obj)/$(datafile_y): $(obj)/gen_init_cpio $(deps_initramfs) klibcdirs
$(Q)$(initramfs) -l $(ramfs-input) > $(obj)/.initramfs_data.cpio.d
$(call if_changed,initfs)does not have FORCE on the end, so I guess is not picking up the dependency properly, perhaps?
Thanks,
Nick
Quoted 257 lines · Nicholas Piggin
Quoted 248 lines · On Thu, 7 Sep 2017 05:50:30 -0700 Florian Fainelli <[email protected]> wrote:
Quoted 100 lines · Nicholas Piggin
Quoted 97 lines · On Mon, 28 Aug 2017 13:03:31 -0700 Florian Fainelli <[email protected]>…
Quoted 70 lines · Nicholas Piggin
Quoted 63 lines · On Sat, 20 May 2017 20:33:35 -0700 Florian Fainelli <[email protected]>…
Commit
db2aa7fd15e8("initramfs: allow again choice of the embedded initram compression algorithm") introduced the possibility to select the initramfs compression algorithm fromKconfigand while this is a nice feature it broke the use case described below.Here is what my build system does:
- kernel is initially configured not to have an initramfs included
- build the user space root file system
- re-configure the kernel to have an initramfs included (CONFIG_INITRAMFS_SOURCE="/path/to/romfs") and set relevantCONFIG_INITRAMFSoptions, in my case, no compression option (CONFIG_INITRAMFS_COMPRESSION_NONE)
- kernel is re-built with these options -> kernel+initramfs image is copied
- kernel is re-built again without these options -> kernel image is copiedBuilding a kernel without an initramfs means setting this option:
CONFIG_INITRAMFS_SOURCE="" (and this one only)whereas building a kernel with an initramfs means setting these options:
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000CONFIG_INITRAMFS_COMPRESSION_NONE=yCONFIG_INITRAMFS_COMPRESSION=""Commit
db2aa7fd15e857891cefbada8348c8d938c7a2bc("initramfs: allow again choice of the embedded initram compression algorithm") is problematic becauseCONFIG_INITRAMFS_COMPRESSIONwhich is used to determine the initramfs_data.cpio extension/compression is a string, and due to howKconfigworks it will evaluate in order, how to assign it.Setting
CONFIG_INITRAMFS_COMPRESSION_NONEwithCONFIG_INITRAMFS_SOURCE="" cannot possibly work (because of the depends on INITRAMFS_SOURCE!="" imposed onCONFIG_INITRAMFS_COMPRESSION) yet we still getCONFIG_INITRAMFS_COMPRESSIONassigned to ".gz" becauseCONFIG_RD_GZIP=yis set in my kernel, even when there is no initramfs being built.So we basically end-up generating two initramfs_data.cpio* files, one without extension, and one with .gz. This causes usr/Makefile to track
usr/initramfs_data.cpio.gz, and notusr/initramfs_data.cpioanymore, that is also largely problematic after9e3596b0c6539e28546ff7c72a06576627068353("kbuild: initramfs cleanup, set target fromKconfig") because we used to track all possibleinitramfs_datafiles in the $(targets) variable before that commit.The end result is that the kernel with an initramfs clearly does not contain what we expect it to, it has a stale initramfs_data.cpio file built into it, and we keep re-generating an initramfs_data.cpio.gz file which is not the one that we want to include in the kernel image proper.
The fix consists in hiding
CONFIG_INITRAMFS_COMPRESSIONwhenCONFIG_INITRAMFS_SOURCE="". This puts us back in a state to the pre-4.10 behavior where we can properly disable and re-enable initramfs within the same kernel .config file, and be in control of whatCONFIG_INITRAMFS_COMPRESSIONis set to.Quoted 3 lines
Fixes db2aa7fd15e8 ("initramfs: allow again choice of the embedded initram compression algorithm")Fixes 9e3596b0c653 ("kbuild: initramfs cleanup, set target from Kconfig")Signed-off-by Florian Fainelli <[email protected]>Quoted 7 lines
This is very thorough, thank you for tracking it down and fixing it.
I can't say I've worked through the problem in the code, but your changelog and the proposed fix seem reasonable to me. So for what it's worth:
Acked-by Nicholas Piggin <[email protected]>Well, I am looking at this again, and it's still broken, the same test case is involved, except this time, I am switching beween no-initramfs and initramfs with gzip compression (the key thing is using a compression of some sort). The end result is the following:
- change stuff in the rootfs
- build the kernel with initramfs,CONFIG_INITRAMFS_COMPRESSION_GZIP=y,usr/initramfs_data.cpio.gzgets generated correctly the first time
- build the kernel without initramfs,CONFIG_INITRAMFS_COMPRESSION_NONE=y,usr/initramfs_data.cpiogets generatedNow back to step 1 add some files, and we can see that
usr/initramfs_data.cpio.gzis now stale from before...So while my earlier fix switched the initramfs w/o compression to no initramfs rebuild, now this does not work because we still have two files left to be tracked:
Quoted 7 lines
Quoted 4 lines
usr/initramfs_data.cpio(no compression, or when initramfs is disabled) and usr/initramfs_data.cpio.$(suffix_y)How would you go about solving this?
I don't see the problem. When I change back to .gz, modify the source directory, and rebuild, it rebuilds a new initramfs_data.cpio.gz here.
This is really puzzling me, so here is what is going on my side:
First time build w/ initramfs enabled:
grep INITRAMFS .config
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000
#CONFIG_INITRAMFS_COMPRESSION_NONEis not setCONFIG_INITRAMFS_COMPRESSION_GZIP=y
#CONFIG_INITRAMFS_COMPRESSION_BZIP2is not set
#CONFIG_INITRAMFS_COMPRESSION_LZMAis not set
#CONFIG_INITRAMFS_COMPRESSION_XZis not set
#CONFIG_INITRAMFS_COMPRESSION_LZOis not set
#CONFIG_INITRAMFS_COMPRESSION_LZ4is not setCONFIG_INITRAMFS_COMPRESSION=".gz"fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 6.8M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:37 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:38 ../(quoted diff) +0
-9-rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:37 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:37 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 9.6K Sep 7 05:36 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37.initramfs_data.cpio.gz.cmd
(quoted diff) +0
-7-rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:37 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:34 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:35 modules.orderNow let's reconfigure this kernel w/o initramfs and rebuild:
grep INITRAMFS .config
CONFIG_INITRAMFS_SOURCE=""fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 3.6M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:44 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:44 ../(quoted diff) +0
-11-rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:44 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:44 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 512 Sep 7 05:44 initramfs_data.cpio -rw-rw-r-- 1 fainelli fainelli 105 Sep 7 05:44 .initramfs_data.cpio.cmd -rw-rw-r-- 1 fainelli fainelli 52 Sep 7 05:44 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37.initramfs_data.cpio.gz.cmd
(quoted diff) +0
-7-rw-rw-r-- 1 fainelli fainelli 1.3K Sep 7 05:44 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:44 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:34 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:35 modules.orderNow let's rebuild with a new application added to the root filesystem:
grep INITRAMFS .config
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000
#CONFIG_INITRAMFS_COMPRESSION_NONEis not setCONFIG_INITRAMFS_COMPRESSION_GZIP=y
#CONFIG_INITRAMFS_COMPRESSION_BZIP2is not set
#CONFIG_INITRAMFS_COMPRESSION_LZMAis not set
#CONFIG_INITRAMFS_COMPRESSION_XZis not set
#CONFIG_INITRAMFS_COMPRESSION_LZOis not set
#CONFIG_INITRAMFS_COMPRESSION_LZ4is not setCONFIG_INITRAMFS_COMPRESSION=".gz"We are not fine here:
scripts/kconfig/conf--silentoldconfigKconfig
## configuration written to .config # CHK include/config/kernel.release CHK include/generated/uapi/linux/version.h 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 AS usr/initramfs_data.o^===== we should have called gzip to re-generate the initramfs, we did not
AR usr/built-in.o GEN .version CHK include/generated/compile.h UPD include/generated/compile.h CC init/version.o AR init/built-in.o AR built-in.o LD vmlinux.oMODPOST vmlinux.o
And the timestamps here confirm that:
fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 6.9M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:46 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:46 ../(quoted diff) +0
-11-rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:46 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:46 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 512 Sep 7 05:44 initramfs_data.cpio -rw-rw-r-- 1 fainelli fainelli 105 Sep 7 05:44 .initramfs_data.cpio.cmd -rw-rw-r-- 1 fainelli fainelli 11K Sep 7 05:46 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37.initramfs_data.cpio.gz.cmd
(quoted diff) +0
-7-rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:46 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:46 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:46 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:46 modules.orderQuoted 3 lines
Worse yet,
usr/.initramfs_data.cpio.ddoes list the binary that was just added, but for some reason, this did not trigger a rebuild of initramfs_data.cpio.gz do you see any reasons why?This rule
Quoted 6 lines
$(obj)/$(datafile_y): $(obj)/gen_init_cpio $(deps_initramfs) klibcdirs
$(Q)$(initramfs) -l $(ramfs-input) > $(obj)/.initramfs_data.cpio.d $(call if_changed,initfs)does not have FORCE on the end, so I guess is not picking up the dependency properly, perhaps?
I tried that:
usr/Makefile +1 -1
@@ -51,6 +51,6 @@ $(deps_initramfs): klibcdirs # 2) There are changes in which files are included (added or deleted) # 3) If gen_init_cpio are newer than initramfs_data.cpio # 4) arguments to gen_initramfs.sh changes -$(obj)/$(datafile_y): $(obj)/gen_init_cpio $(deps_initramfs) klibcdirs +$(obj)/$(datafile_y): $(obj)/gen_init_cpio $(deps_initramfs) klibcdirs
FORCE
$(Q)$(initramfs) -l $(ramfs-input) > $(obj)/.initramfs_data.cpio.d
$(call if_changed,initfbut this did not change anything.
We are re-generating .initramfs_data.cpio.d and its timestamp is clearly updated so it's got to be cmd_initfs which does not get invoked.
.initramfs_data.cpio.gz.cmd is correct:
cmd_usr/initramfs_data.cpio.gz := /bin/bash ./scripts/gen_initramfs_list.sh -o usr/initramfs_data.cpio.gz -u 1000 -g 1000 /home/fainelli/work/uclinux-rootfs/romfs /home/fainelli/work/uclinux-rootfs/misc/initramfs.dev
and was generated the first time we did generate the gzip initramfs, so the command has not changed, nor its arguments, so we just don't call it.
I did try to use if_changed_dep, which rightfully pointed that we don't have a .initramfs_data.cpio.gz.d, but this gets us very close from the only thing that I could get working thus far. Another way to possibly solve it is to void generating a cmd_$($@) and use the same name irrespective of the compression type used, that should force a change everytime...
Thanks Nick!
usr/Makefile +1 -0
@@ -7,6 +7,7 @@ PHONY += klibcdirs suffix_y = $(subst $\",,$(CONFIG_INITRAMFS_COMPRESSION)) datafile_y = initramfs_data.cpio$(suffix_y) +datafile_d_y = .$(datafile_y).d AFLAGS_initramfs_data.o += -DINITRAMFS_IMAGE="usr/$(datafile_y)"
@@ -29,12 +30,12 @@ ramfs-args := \ $(if $(CONFIG_INITRAMFS_ROOT_UID), -u
$(CONFIG_INITRAMFS_ROOT_UID)) \
$(if $(CONFIG_INITRAMFS_ROOT_GID), -g $(CONFIG_INITRAMFS_ROOT_GID))-# .initramfs_data.cpio.d is used to identify all files included +# $(datafile_d_y) is used to identify all files included
# in initramfs and to detect if any files are added/removed. # Removed files are identified by directory timestamp being updated # The dependency list is generated by gen_initramfs.sh -l -ifneq ($(wildcard $(obj)/.initramfs_data.cpio.d),) - include $(obj)/.initramfs_data.cpio.d +ifneq ($(wildcard $(obj)/$(datafile_d_y)),) + include $(obj)/$(datafile_d_y) endif
quiet_cmd_initfs = GEN $@
@@ -52,5 +53,5 @@ $(deps_initramfs): klibcdirs
# 3) If gen_init_cpio are newer than initramfs_data.cpio
# 4) arguments to gen_initramfs.sh changes $(obj)/$(datafile_y): $(obj)/gen_init_cpio $(deps_initramfs) klibcdirs
- $(Q)$(initramfs) -l $(ramfs-input) > $(obj)/.initramfs_data.cpio.d
+ $(Q)$(initramfs) -l $(ramfs-input) > $(obj)/$(datafile_d_y)
$(call if_changed,initfs)
Signature · 1 line
Florian
Quoted 298 lines · Florian Fainelli
Quoted 257 lines · Nicholas Piggin
Quoted 248 lines · On Thu, 7 Sep 2017 05:50:30 -0700 Florian Fainelli <[email protected]> wrote:
Quoted 100 lines · Nicholas Piggin
Quoted 97 lines · On Mon, 28 Aug 2017 13:03:31 -0700 Florian Fainelli <[email protected]>…
Quoted 70 lines · Nicholas Piggin
Quoted 63 lines · On Sat, 20 May 2017 20:33:35 -0700 Florian Fainelli <[email protected]>…
Commit
db2aa7fd15e8("initramfs: allow again choice of the embedded initram compression algorithm") introduced the possibility to select the initramfs compression algorithm fromKconfigand while this is a nice feature it broke the use case described below.Here is what my build system does:
- kernel is initially configured not to have an initramfs included
- build the user space root file system
- re-configure the kernel to have an initramfs included (CONFIG_INITRAMFS_SOURCE="/path/to/romfs") and set relevantCONFIG_INITRAMFSoptions, in my case, no compression option (CONFIG_INITRAMFS_COMPRESSION_NONE)
- kernel is re-built with these options -> kernel+initramfs image is copied
- kernel is re-built again without these options -> kernel image is copiedBuilding a kernel without an initramfs means setting this option:
CONFIG_INITRAMFS_SOURCE="" (and this one only)whereas building a kernel with an initramfs means setting these options:
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000CONFIG_INITRAMFS_COMPRESSION_NONE=yCONFIG_INITRAMFS_COMPRESSION=""Commit
db2aa7fd15e857891cefbada8348c8d938c7a2bc("initramfs: allow again choice of the embedded initram compression algorithm") is problematic becauseCONFIG_INITRAMFS_COMPRESSIONwhich is used to determine the initramfs_data.cpio extension/compression is a string, and due to howKconfigworks it will evaluate in order, how to assign it.Setting
CONFIG_INITRAMFS_COMPRESSION_NONEwithCONFIG_INITRAMFS_SOURCE="" cannot possibly work (because of the depends on INITRAMFS_SOURCE!="" imposed onCONFIG_INITRAMFS_COMPRESSION) yet we still getCONFIG_INITRAMFS_COMPRESSIONassigned to ".gz" becauseCONFIG_RD_GZIP=yis set in my kernel, even when there is no initramfs being built.So we basically end-up generating two initramfs_data.cpio* files, one without extension, and one with .gz. This causes usr/Makefile to track
usr/initramfs_data.cpio.gz, and notusr/initramfs_data.cpioanymore, that is also largely problematic after9e3596b0c6539e28546ff7c72a06576627068353("kbuild: initramfs cleanup, set target fromKconfig") because we used to track all possibleinitramfs_datafiles in the $(targets) variable before that commit.The end result is that the kernel with an initramfs clearly does not contain what we expect it to, it has a stale initramfs_data.cpio file built into it, and we keep re-generating an initramfs_data.cpio.gz file which is not the one that we want to include in the kernel image proper.
The fix consists in hiding
CONFIG_INITRAMFS_COMPRESSIONwhenCONFIG_INITRAMFS_SOURCE="". This puts us back in a state to the pre-4.10 behavior where we can properly disable and re-enable initramfs within the same kernel .config file, and be in control of whatCONFIG_INITRAMFS_COMPRESSIONis set to.Quoted 3 lines
Fixes db2aa7fd15e8 ("initramfs: allow again choice of the embedded initram compression algorithm")Fixes 9e3596b0c653 ("kbuild: initramfs cleanup, set target from Kconfig")Signed-off-by Florian Fainelli <[email protected]>Quoted 7 lines
This is very thorough, thank you for tracking it down and fixing it.
I can't say I've worked through the problem in the code, but your changelog and the proposed fix seem reasonable to me. So for what it's worth:
Acked-by Nicholas Piggin <[email protected]>Well, I am looking at this again, and it's still broken, the same test case is involved, except this time, I am switching beween no-initramfs and initramfs with gzip compression (the key thing is using a compression of some sort). The end result is the following:
- change stuff in the rootfs
- build the kernel with initramfs,CONFIG_INITRAMFS_COMPRESSION_GZIP=y,usr/initramfs_data.cpio.gzgets generated correctly the first time
- build the kernel without initramfs,CONFIG_INITRAMFS_COMPRESSION_NONE=y,usr/initramfs_data.cpiogets generatedNow back to step 1 add some files, and we can see that
usr/initramfs_data.cpio.gzis now stale from before...So while my earlier fix switched the initramfs w/o compression to no initramfs rebuild, now this does not work because we still have two files left to be tracked:
Quoted 7 lines
Quoted 4 lines
usr/initramfs_data.cpio(no compression, or when initramfs is disabled) and usr/initramfs_data.cpio.$(suffix_y)How would you go about solving this?
I don't see the problem. When I change back to .gz, modify the source directory, and rebuild, it rebuilds a new initramfs_data.cpio.gz here.
This is really puzzling me, so here is what is going on my side:
First time build w/ initramfs enabled:
grep INITRAMFS .config
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000
#CONFIG_INITRAMFS_COMPRESSION_NONEis not setCONFIG_INITRAMFS_COMPRESSION_GZIP=y
#CONFIG_INITRAMFS_COMPRESSION_BZIP2is not set
#CONFIG_INITRAMFS_COMPRESSION_LZMAis not set
#CONFIG_INITRAMFS_COMPRESSION_XZis not set
#CONFIG_INITRAMFS_COMPRESSION_LZOis not set
#CONFIG_INITRAMFS_COMPRESSION_LZ4is not setCONFIG_INITRAMFS_COMPRESSION=".gz"fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 6.8M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:37 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:38 ../(quoted diff) +0
-9-rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:37 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:37 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 9.6K Sep 7 05:36 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37.initramfs_data.cpio.gz.cmd
(quoted diff) +0
-7-rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:37 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:34 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:35 modules.orderNow let's reconfigure this kernel w/o initramfs and rebuild:
grep INITRAMFS .config
CONFIG_INITRAMFS_SOURCE=""fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 3.6M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:44 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:44 ../(quoted diff) +0
-11-rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:44 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:44 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 512 Sep 7 05:44 initramfs_data.cpio -rw-rw-r-- 1 fainelli fainelli 105 Sep 7 05:44 .initramfs_data.cpio.cmd -rw-rw-r-- 1 fainelli fainelli 52 Sep 7 05:44 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37.initramfs_data.cpio.gz.cmd
(quoted diff) +0
-7-rw-rw-r-- 1 fainelli fainelli 1.3K Sep 7 05:44 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:44 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:34 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:35 modules.orderNow let's rebuild with a new application added to the root filesystem:
grep INITRAMFS .config
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000
#CONFIG_INITRAMFS_COMPRESSION_NONEis not setCONFIG_INITRAMFS_COMPRESSION_GZIP=y
#CONFIG_INITRAMFS_COMPRESSION_BZIP2is not set
#CONFIG_INITRAMFS_COMPRESSION_LZMAis not set
#CONFIG_INITRAMFS_COMPRESSION_XZis not set
#CONFIG_INITRAMFS_COMPRESSION_LZOis not set
#CONFIG_INITRAMFS_COMPRESSION_LZ4is not setCONFIG_INITRAMFS_COMPRESSION=".gz"We are not fine here:
scripts/kconfig/conf--silentoldconfigKconfig
## configuration written to .config # CHK include/config/kernel.release CHK include/generated/uapi/linux/version.h 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 AS usr/initramfs_data.o^===== we should have called gzip to re-generate the initramfs, we did not
AR usr/built-in.o GEN .version CHK include/generated/compile.h UPD include/generated/compile.h CC init/version.o AR init/built-in.o AR built-in.o LD vmlinux.oMODPOST vmlinux.o
And the timestamps here confirm that:
fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 6.9M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:46 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:46 ../(quoted diff) +0
-11-rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:46 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:46 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 512 Sep 7 05:44 initramfs_data.cpio -rw-rw-r-- 1 fainelli fainelli 105 Sep 7 05:44 .initramfs_data.cpio.cmd -rw-rw-r-- 1 fainelli fainelli 11K Sep 7 05:46 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37.initramfs_data.cpio.gz.cmd
(quoted diff) +0
-7-rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:46 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:46 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:46 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:46 modules.orderQuoted 3 lines
Worse yet,
usr/.initramfs_data.cpio.ddoes list the binary that was just added, but for some reason, this did not trigger a rebuild of initramfs_data.cpio.gz do you see any reasons why?This rule
Quoted 6 lines
$(obj)/$(datafile_y): $(obj)/gen_init_cpio $(deps_initramfs) klibcdirs
$(Q)$(initramfs) -l $(ramfs-input) > $(obj)/.initramfs_data.cpio.d $(call if_changed,initfs)does not have FORCE on the end, so I guess is not picking up the dependency properly, perhaps?
I tried that:
usr/Makefile +1
-1@@ -51,6 +51,6 @@ $(deps_initramfs): klibcdirs # 2) There are changes in which files are included (added or deleted) # 3) If gen_init_cpio are newer than initramfs_data.cpio # 4) arguments to gen_initramfs.sh changes -$(obj)/$(datafile_y): $(obj)/gen_init_cpio $(deps_initramfs) klibcdirs +$(obj)/$(datafile_y): $(obj)/gen_init_cpio $(deps_initramfs) klibcdirsFORCE
$(Q)$(initramfs) -l $(ramfs-input) > $(obj)/.initramfs_data.cpio.d $(call if_changed,initfbut this did not change anything.
We are re-generating .initramfs_data.cpio.d and its timestamp is clearly updated so it's got to be
cmd_initfswhich does not get invoked..initramfs_data.cpio.gz.cmd is correct:
cmd_usr/initramfs_data.cpio.gz :=
/bin/bash./scripts/gen_initramfs_list.sh-ousr/initramfs_data.cpio.gz-u 1000 -g 1000/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.devand was generated the first time we did generate the gzip initramfs, so the command has not changed, nor its arguments, so we just don't call it.
Quoted 8 lines
I did try to use
if_changed_dep, which rightfully pointed that we don't have a .initramfs_data.cpio.gz.d, but this gets us very close from the only thing that I could get working thus far. Another way to possibly solve it is to void generating a cmd_$($@) and use the same name irrespective of the compression type used, that should force a change everytime...Thanks Nick!
Nick, is the patch below acceptable to you?
Quoted 38 lines
usr/Makefile +4
-3@@ -7,6 +7,7 @@ PHONY += klibcdirs suffix_y = $(subst $\",,$(CONFIG_INITRAMFS_COMPRESSION)) datafile_y = initramfs_data.cpio$(suffix_y) +datafile_d_y = .$(datafile_y).d AFLAGS_initramfs_data.o += -DINITRAMFS_IMAGE="usr/$(datafile_y)"@@ -29,12 +30,12 @@ ramfs-args := \ $(if $(CONFIG_INITRAMFS_ROOT_UID), -u $(CONFIG_INITRAMFS_ROOT_UID)) \ $(if $(CONFIG_INITRAMFS_ROOT_GID), -g $(CONFIG_INITRAMFS_ROOT_GID)) -# .initramfs_data.cpio.d is used to identify all files included +# $(datafile_d_y) is used to identify all files included # in initramfs and to detect if any files are added/removed. # Removed files are identified by directory timestamp being updated # The dependency list is generated by gen_initramfs.sh -l -ifneq ($(wildcard $(obj)/.initramfs_data.cpio.d),) - include $(obj)/.initramfs_data.cpio.d +ifneq ($(wildcard $(obj)/$(datafile_d_y)),) + include $(obj)/$(datafile_d_y) endif
quiet_cmd_initfs= GEN $@usr/Makefile +1
-1@@ -52,5 +53,5 @@ $(deps_initramfs): klibcdirs # 3) If gen_init_cpio are newer than initramfs_data.cpio # 4) arguments to gen_initramfs.sh changes $(obj)/$(datafile_y): $(obj)/gen_init_cpio $(deps_initramfs) klibcdirs - $(Q)$(initramfs) -l $(ramfs-input) > $(obj)/.initramfs_data.cpio.d + $(Q)$(initramfs) -l $(ramfs-input) > $(obj)/$(datafile_d_y) $(call if_changed,initfs)
Signature · 1 line
Florian
Quoted 306 lines · On Thu, 14 Sep 2017 15:59:50 -0700 Florian Fainelli <[email protected]>…
Quoted 305 lines · Florian Fainelli
Quoted 257 lines · Nicholas Piggin
Quoted 248 lines · On Thu, 7 Sep 2017 05:50:30 -0700 Florian Fainelli <[email protected]> wrote:
Quoted 100 lines · Nicholas Piggin
Quoted 97 lines · On Mon, 28 Aug 2017 13:03:31 -0700 Florian Fainelli <[email protected]>…
Quoted 70 lines · Nicholas Piggin
Quoted 63 lines · On Sat, 20 May 2017 20:33:35 -0700 Florian Fainelli <[email protected]>…
Commit
db2aa7fd15e8("initramfs: allow again choice of the embedded initram compression algorithm") introduced the possibility to select the initramfs compression algorithm fromKconfigand while this is a nice feature it broke the use case described below.Here is what my build system does:
- kernel is initially configured not to have an initramfs included
- build the user space root file system
- re-configure the kernel to have an initramfs included (CONFIG_INITRAMFS_SOURCE="/path/to/romfs") and set relevantCONFIG_INITRAMFSoptions, in my case, no compression option (CONFIG_INITRAMFS_COMPRESSION_NONE)
- kernel is re-built with these options -> kernel+initramfs image is copied
- kernel is re-built again without these options -> kernel image is copiedBuilding a kernel without an initramfs means setting this option:
CONFIG_INITRAMFS_SOURCE="" (and this one only)whereas building a kernel with an initramfs means setting these options:
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000CONFIG_INITRAMFS_COMPRESSION_NONE=yCONFIG_INITRAMFS_COMPRESSION=""Commit
db2aa7fd15e857891cefbada8348c8d938c7a2bc("initramfs: allow again choice of the embedded initram compression algorithm") is problematic becauseCONFIG_INITRAMFS_COMPRESSIONwhich is used to determine the initramfs_data.cpio extension/compression is a string, and due to howKconfigworks it will evaluate in order, how to assign it.Setting
CONFIG_INITRAMFS_COMPRESSION_NONEwithCONFIG_INITRAMFS_SOURCE="" cannot possibly work (because of the depends on INITRAMFS_SOURCE!="" imposed onCONFIG_INITRAMFS_COMPRESSION) yet we still getCONFIG_INITRAMFS_COMPRESSIONassigned to ".gz" becauseCONFIG_RD_GZIP=yis set in my kernel, even when there is no initramfs being built.So we basically end-up generating two initramfs_data.cpio* files, one without extension, and one with .gz. This causes usr/Makefile to track
usr/initramfs_data.cpio.gz, and notusr/initramfs_data.cpioanymore, that is also largely problematic after9e3596b0c6539e28546ff7c72a06576627068353("kbuild: initramfs cleanup, set target fromKconfig") because we used to track all possibleinitramfs_datafiles in the $(targets) variable before that commit.The end result is that the kernel with an initramfs clearly does not contain what we expect it to, it has a stale initramfs_data.cpio file built into it, and we keep re-generating an initramfs_data.cpio.gz file which is not the one that we want to include in the kernel image proper.
The fix consists in hiding
CONFIG_INITRAMFS_COMPRESSIONwhenCONFIG_INITRAMFS_SOURCE="". This puts us back in a state to the pre-4.10 behavior where we can properly disable and re-enable initramfs within the same kernel .config file, and be in control of whatCONFIG_INITRAMFS_COMPRESSIONis set to.Quoted 3 lines
Fixes db2aa7fd15e8 ("initramfs: allow again choice of the embedded initram compression algorithm")Fixes 9e3596b0c653 ("kbuild: initramfs cleanup, set target from Kconfig")Signed-off-by Florian Fainelli <[email protected]>Quoted 7 lines
This is very thorough, thank you for tracking it down and fixing it.
I can't say I've worked through the problem in the code, but your changelog and the proposed fix seem reasonable to me. So for what it's worth:
Acked-by Nicholas Piggin <[email protected]>Well, I am looking at this again, and it's still broken, the same test case is involved, except this time, I am switching beween no-initramfs and initramfs with gzip compression (the key thing is using a compression of some sort). The end result is the following:
- change stuff in the rootfs
- build the kernel with initramfs,CONFIG_INITRAMFS_COMPRESSION_GZIP=y,usr/initramfs_data.cpio.gzgets generated correctly the first time
- build the kernel without initramfs,CONFIG_INITRAMFS_COMPRESSION_NONE=y,usr/initramfs_data.cpiogets generatedNow back to step 1 add some files, and we can see that
usr/initramfs_data.cpio.gzis now stale from before...So while my earlier fix switched the initramfs w/o compression to no initramfs rebuild, now this does not work because we still have two files left to be tracked:
Quoted 7 lines
Quoted 4 lines
usr/initramfs_data.cpio(no compression, or when initramfs is disabled) and usr/initramfs_data.cpio.$(suffix_y)How would you go about solving this?
I don't see the problem. When I change back to .gz, modify the source directory, and rebuild, it rebuilds a new initramfs_data.cpio.gz here.
This is really puzzling me, so here is what is going on my side:
First time build w/ initramfs enabled:
grep INITRAMFS .config
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000
#CONFIG_INITRAMFS_COMPRESSION_NONEis not setCONFIG_INITRAMFS_COMPRESSION_GZIP=y
#CONFIG_INITRAMFS_COMPRESSION_BZIP2is not set
#CONFIG_INITRAMFS_COMPRESSION_LZMAis not set
#CONFIG_INITRAMFS_COMPRESSION_XZis not set
#CONFIG_INITRAMFS_COMPRESSION_LZOis not set
#CONFIG_INITRAMFS_COMPRESSION_LZ4is not setCONFIG_INITRAMFS_COMPRESSION=".gz"fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 6.8M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:37 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:38 ../(quoted diff) +0
-9-rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:37 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:37 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 9.6K Sep 7 05:36 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37.initramfs_data.cpio.gz.cmd
(quoted diff) +0
-7-rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:37 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:34 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:35 modules.orderNow let's reconfigure this kernel w/o initramfs and rebuild:
grep INITRAMFS .config
CONFIG_INITRAMFS_SOURCE=""fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 3.6M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:44 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:44 ../(quoted diff) +0
-11-rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:44 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:44 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 512 Sep 7 05:44 initramfs_data.cpio -rw-rw-r-- 1 fainelli fainelli 105 Sep 7 05:44 .initramfs_data.cpio.cmd -rw-rw-r-- 1 fainelli fainelli 52 Sep 7 05:44 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37.initramfs_data.cpio.gz.cmd
(quoted diff) +0
-7-rw-rw-r-- 1 fainelli fainelli 1.3K Sep 7 05:44 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:44 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:34 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:35 modules.orderNow let's rebuild with a new application added to the root filesystem:
grep INITRAMFS .config
CONFIG_INITRAMFS_SOURCE="/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.dev"CONFIG_INITRAMFS_ROOT_UID=1000CONFIG_INITRAMFS_ROOT_GID=1000
#CONFIG_INITRAMFS_COMPRESSION_NONEis not setCONFIG_INITRAMFS_COMPRESSION_GZIP=y
#CONFIG_INITRAMFS_COMPRESSION_BZIP2is not set
#CONFIG_INITRAMFS_COMPRESSION_LZMAis not set
#CONFIG_INITRAMFS_COMPRESSION_XZis not set
#CONFIG_INITRAMFS_COMPRESSION_LZOis not set
#CONFIG_INITRAMFS_COMPRESSION_LZ4is not setCONFIG_INITRAMFS_COMPRESSION=".gz"We are not fine here:
scripts/kconfig/conf--silentoldconfigKconfig
## configuration written to .config # CHK include/config/kernel.release CHK include/generated/uapi/linux/version.h 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 AS usr/initramfs_data.o^===== we should have called gzip to re-generate the initramfs, we did not
AR usr/built-in.o GEN .version CHK include/generated/compile.h UPD include/generated/compile.h CC init/version.o AR init/built-in.o AR built-in.o LD vmlinux.oMODPOST vmlinux.o
And the timestamps here confirm that:
fainelli@fainelli-laptop:[~/../linux]$ ls -la usr total 6.9M drwxrwxr-x 2 fainelli fainelli 4.0K Sep 7 05:46 ./ drwxrwxr-x 27 fainelli fainelli 16K Sep 7 05:46 ../(quoted diff) +0
-11-rw-rw-r-- 1 fainelli fainelli 146 Sep 7 05:46 built-in.o -rw-rw-r-- 1 fainelli fainelli 112 Sep 7 05:46 .built-in.o.cmd -rwxrwxr-x 1 fainelli fainelli 23K Sep 7 05:36 gen_init_cpio* -rw-rw-r-- 1 fainelli fainelli 13K Oct 13 2016 gen_init_cpio.c -rw-rw-r-- 1 fainelli fainelli 3.3K Sep 7 05:36 .gen_init_cpio.cmd -rw-rw-r-- 1 fainelli fainelli 151 Aug 28 2016 .gitignore -rw-rw-r-- 1 fainelli fainelli 512 Sep 7 05:44 initramfs_data.cpio -rw-rw-r-- 1 fainelli fainelli 105 Sep 7 05:44 .initramfs_data.cpio.cmd -rw-rw-r-- 1 fainelli fainelli 11K Sep 7 05:46 .initramfs_data.cpio.d -rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:37 initramfs_data.cpio.gz -rw-rw-r-- 1 fainelli fainelli 220 Sep 7 05:37.initramfs_data.cpio.gz.cmd
(quoted diff) +0
-7-rw-rw-r-- 1 fainelli fainelli 3.3M Sep 7 05:46 initramfs_data.o -rw-rw-r-- 1 fainelli fainelli 2.7K Sep 7 05:46 .initramfs_data.o.cmd -rw-rw-r-- 1 fainelli fainelli 1.3K Aug 28 2016 initramfs_data.S -rw-rw-r-- 1 fainelli fainelli 7.8K Aug 18 19:25 Kconfig -rw-rw-r-- 1 fainelli fainelli 2.0K Aug 31 20:46 Makefile -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:46 modules.builtin -rw-rw-r-- 1 fainelli fainelli 0 Sep 7 05:46 modules.orderQuoted 3 lines
Worse yet,
usr/.initramfs_data.cpio.ddoes list the binary that was just added, but for some reason, this did not trigger a rebuild of initramfs_data.cpio.gz do you see any reasons why?This rule
Quoted 6 lines
$(obj)/$(datafile_y): $(obj)/gen_init_cpio $(deps_initramfs) klibcdirs
$(Q)$(initramfs) -l $(ramfs-input) > $(obj)/.initramfs_data.cpio.d $(call if_changed,initfs)does not have FORCE on the end, so I guess is not picking up the dependency properly, perhaps?
I tried that:
usr/Makefile +1
-1@@ -51,6 +51,6 @@ $(deps_initramfs): klibcdirs # 2) There are changes in which files are included (added or deleted) # 3) If gen_init_cpio are newer than initramfs_data.cpio # 4) arguments to gen_initramfs.sh changes -$(obj)/$(datafile_y): $(obj)/gen_init_cpio $(deps_initramfs) klibcdirs +$(obj)/$(datafile_y): $(obj)/gen_init_cpio $(deps_initramfs) klibcdirsFORCE
$(Q)$(initramfs) -l $(ramfs-input) > $(obj)/.initramfs_data.cpio.d $(call if_changed,initfbut this did not change anything.
We are re-generating .initramfs_data.cpio.d and its timestamp is clearly updated so it's got to be
cmd_initfswhich does not get invoked..initramfs_data.cpio.gz.cmd is correct:
cmd_usr/initramfs_data.cpio.gz :=
/bin/bash./scripts/gen_initramfs_list.sh-ousr/initramfs_data.cpio.gz-u 1000 -g 1000/home/fainelli/work/uclinux-rootfs/romfs/home/fainelli/work/uclinux-rootfs/misc/initramfs.devand was generated the first time we did generate the gzip initramfs, so the command has not changed, nor its arguments, so we just don't call it.
I did try to use
if_changed_dep, which rightfully pointed that we don't have a .initramfs_data.cpio.gz.d, but this gets us very close from the only thing that I could get working thus far. Another way to possibly solve it is to void generating a cmd_$($@) and use the same name irrespective of the compression type used, that should force a change everytime...
Quoted 3 lines
Quoted 1 line
Thanks Nick!
Nick, is the patch below acceptable to you?
Well it looks reasonable, and it fixes your problem. I'm not maintainer of this code or a Makefile wizard though :) Perhaps send it up to the linux-kbuild maintainers?
Thanks,
Nick