[U-Boot] fit image: boot gzipped ramdisk does not work anymore
Hello Julius,
since your commit (just gone into mainline):
b1307f884a91 ("fit: Support compression for non-kernel components (e.g. FDT)")
I see on an imx6ull based board when booting (FIT image with kernel, dtb and initramdisk), fit image source created from yocto, see [1]:
## Loading kernel from FIT Image at 84000000 ... Using 'conf@imx6ull-k30rf.dtb' configuration Trying 'kernel@1' kernel subimage Description: Linux kernel Type: Kernel Image Compression: uncompressed Data Start: 0x840000e8 Data Size: 5154568 Bytes = 4.9 MiB Architecture: ARM OS: Linux Load Address: 0x82000000 Entry Point: 0x82000000 Hash algo: sha1 Hash value: a3ba8a09b197c0d1acab834c69c5aea677d4a092 Verifying Hash Integrity ... sha1+ OK ## Loading ramdisk from FIT Image at 84000000 ... Using 'conf@imx6ull-k30rf.dtb' configuration Trying 'ramdisk@1' ramdisk subimage Description: k30rf-image-initramfs Type: RAMDisk Image Compression: gzip compressed Data Start: 0x844f0cf8 Data Size: 6769496 Bytes = 6.5 MiB Architecture: ARM OS: Linux Load Address: unavailable Entry Point: unavailable Hash algo: sha1 Hash value: 507d8b20b18f9bca0e9178c48f634df5da720f65 Verifying Hash Integrity ... sha1+ OK Uncompressing RAMDisk Image Error: Bad gzipped data Error decompressing ramdisk Ramdisk image is corrupt or invalid =>
I found your commit with "git bisect":
hs@xmglap:u-boot [(kein Branch, binäre Suche begonnen bei master)] $ git bisect good b1307f884a913f52a201491053b6d221c4204f60 is the first bad commit commit b1307f884a913f52a201491053b6d221c4204f60 Author: Julius Werner jwerner@chromium.org Date: Wed Jul 24 19:37:55 2019 -0700
fit: Support compression for non-kernel components (e.g. FDT)
This patch adds support for compressing non-kernel image nodes in a FIT image (kernel nodes could already be compressed previously). This can reduce the size of FIT images and therefore improve boot times (especially when an image bundles many different kernel FDTs). The images will automatically be decompressed on load.
This patch does not support extracting compatible strings from compressed FDTs, so it's not very helpful in conjunction with CONFIG_FIT_BEST_MATCH yet, but it can already be used in environments that select the configuration to load explicitly.
Signed-off-by: Julius Werner jwerner@chromium.org Reviewed-by: Simon Glass sjg@chromium.org Reviewed-by: Simon Goldschmidt simon.k.r.goldschmidt@gmail.com
:040000 040000 6280a41241c1c25a28845fa526c5c958e448e347 e3f8ed43853541f868676f077924e7ab242d4da6 M common :040000 040000 406b0c50c03633bad9d4a576d5662f4fa483fc6f 601624cee600c1c06640858519d93188cc312544 M test hs@xmglap:u-boot [(kein Branch, binäre Suche begonnen bei master)] $ git bisect log git bisect start # bad: [bbaf56eda0e63d6d28fbccae0f112ef88203eb4d] Merge tag 'u-boot-amlogic-20190731' of https://gitlab.denx.de/u-boot/custodians/u-boot-amlogic git bisect bad bbaf56eda0e63d6d28fbccae0f112ef88203eb4d # good: [a9aa4c5700c68c070d63a391b51ea8d341b6e8a6] Merge branch '2019-07-24-master-imports' git bisect good a9aa4c5700c68c070d63a391b51ea8d341b6e8a6 # good: [75551c8bfc9545e31ec2ce238cac3857904007b8] Merge branch '2019-07-26-ti-imports' git bisect good 75551c8bfc9545e31ec2ce238cac3857904007b8 # bad: [333755ef7b6f824366eed37ae068c20a4f25a123] Merge branch '2019-07-29-ti-imports' git bisect bad 333755ef7b6f824366eed37ae068c20a4f25a123 # good: [26008cd42b590dc71ee9c1ca667a218542aab342] rockchip: rv1108: Migrate to use common board file git bisect good 26008cd42b590dc71ee9c1ca667a218542aab342 # good: [92430b8fc8aac3b4ab92e9ca8a09d83c4788c609] Merge https://gitlab.denx.de/u-boot/custodians/u-boot-socfpga git bisect good 92430b8fc8aac3b4ab92e9ca8a09d83c4788c609 # bad: [2d64a0f7e952f54375702fb2b854461e402ded9d] Merge branch '2019-07-29-master-imports' git bisect bad 2d64a0f7e952f54375702fb2b854461e402ded9d # bad: [49b10cb4926285b856b207c1f5bb40c75487f08b] gpio: fixes for gpio-hog support git bisect bad 49b10cb4926285b856b207c1f5bb40c75487f08b # bad: [e35d533c79c75e9ad68a2e78da52c8ce58e6ce0d] tools/logos: remove black background of U-Boot logo git bisect bad e35d533c79c75e9ad68a2e78da52c8ce58e6ce0d # bad: [b1307f884a913f52a201491053b6d221c4204f60] fit: Support compression for non-kernel components (e.g. FDT) git bisect bad b1307f884a913f52a201491053b6d221c4204f60 # good: [2090854cd2f228bab546f2718ccdbe1664830d3c] common: Move bootm_decomp_image() to image.c (as image_decomp()) git bisect good 2090854cd2f228bab546f2718ccdbe1664830d3c # first bad commit: [b1307f884a913f52a201491053b6d221c4204f60] fit: Support compression for non-kernel components (e.g. FDT) hs@xmglap:u-boot [(kein Branch, binäre Suche begonnen bei master)] $
base was current master: * bbaf56eda0 - Merge tag 'u-boot-amlogic-20190731' of https://gitlab.denx.de/u-boot/custodians/u-boot-amlogic
I am currently not in my office to dig deeper into it ... may you have an idea?
Thanks!
bye, Heiko
[1] fit image source (created from yocto) /dts-v1/;
/ { description = "U-Boot fitImage for Poky (Yocto Project Reference Distro)/4.19.59/k30rf"; #address-cells = <1>;
images { [...] ramdisk@1 { description = "k30rf-image-initramfs"; data = /incbin/("/work/hs/tbot2go/k30rf/repo/k30rf/build_k30rf/tmp/deploy/images/k30rf/k30rf-image-initramfs-k30rf.cpio.gz"); type = "ramdisk"; arch = "arm"; os = "linux"; compression = "gzip";
Am 01.08.2019 um 10:06 schrieb Heiko Schocher:
Hello Julius,
since your commit (just gone into mainline):
b1307f884a91 ("fit: Support compression for non-kernel components (e.g. FDT)")
I see on an imx6ull based board when booting (FIT image with kernel, dtb and initramdisk), fit image source created from yocto, see [1]:
## Loading kernel from FIT Image at 84000000 ... Using 'conf@imx6ull-k30rf.dtb' configuration Trying 'kernel@1' kernel subimage Description: Linux kernel Type: Kernel Image Compression: uncompressed Data Start: 0x840000e8 Data Size: 5154568 Bytes = 4.9 MiB Architecture: ARM OS: Linux Load Address: 0x82000000 Entry Point: 0x82000000 Hash algo: sha1 Hash value: a3ba8a09b197c0d1acab834c69c5aea677d4a092 Verifying Hash Integrity ... sha1+ OK ## Loading ramdisk from FIT Image at 84000000 ... Using 'conf@imx6ull-k30rf.dtb' configuration Trying 'ramdisk@1' ramdisk subimage Description: k30rf-image-initramfs Type: RAMDisk Image Compression: gzip compressed Data Start: 0x844f0cf8 Data Size: 6769496 Bytes = 6.5 MiB Architecture: ARM OS: Linux Load Address: unavailable Entry Point: unavailable Hash algo: sha1 Hash value: 507d8b20b18f9bca0e9178c48f634df5da720f65 Verifying Hash Integrity ... sha1+ OK Uncompressing RAMDisk Image Error: Bad gzipped data Error decompressing ramdisk Ramdisk image is corrupt or invalid =>
I found your commit with "git bisect":
hs@xmglap:u-boot [(kein Branch, binäre Suche begonnen bei master)] $ git bisect good b1307f884a913f52a201491053b6d221c4204f60 is the first bad commit commit b1307f884a913f52a201491053b6d221c4204f60 Author: Julius Werner jwerner@chromium.org Date: Wed Jul 24 19:37:55 2019 -0700
fit: Support compression for non-kernel components (e.g. FDT) This patch adds support for compressing non-kernel image nodes in a FIT image (kernel nodes could already be compressed previously). This can reduce the size of FIT images and therefore improve boot times (especially when an image bundles many different kernel FDTs). The images will automatically be decompressed on load. This patch does not support extracting compatible strings from compressed FDTs, so it's not very helpful in conjunction with CONFIG_FIT_BEST_MATCH yet, but it can already be used in environments that select the configuration to load explicitly. Signed-off-by: Julius Werner <jwerner@chromium.org> Reviewed-by: Simon Glass <sjg@chromium.org> Reviewed-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>:040000 040000 6280a41241c1c25a28845fa526c5c958e448e347 e3f8ed43853541f868676f077924e7ab242d4da6 M common :040000 040000 406b0c50c03633bad9d4a576d5662f4fa483fc6f 601624cee600c1c06640858519d93188cc312544 M test hs@xmglap:u-boot [(kein Branch, binäre Suche begonnen bei master)] $ git bisect log git bisect start # bad: [bbaf56eda0e63d6d28fbccae0f112ef88203eb4d] Merge tag 'u-boot-amlogic-20190731' of https://gitlab.denx.de/u-boot/custodians/u-boot-amlogic git bisect bad bbaf56eda0e63d6d28fbccae0f112ef88203eb4d # good: [a9aa4c5700c68c070d63a391b51ea8d341b6e8a6] Merge branch '2019-07-24-master-imports' git bisect good a9aa4c5700c68c070d63a391b51ea8d341b6e8a6 # good: [75551c8bfc9545e31ec2ce238cac3857904007b8] Merge branch '2019-07-26-ti-imports' git bisect good 75551c8bfc9545e31ec2ce238cac3857904007b8 # bad: [333755ef7b6f824366eed37ae068c20a4f25a123] Merge branch '2019-07-29-ti-imports' git bisect bad 333755ef7b6f824366eed37ae068c20a4f25a123 # good: [26008cd42b590dc71ee9c1ca667a218542aab342] rockchip: rv1108: Migrate to use common board file git bisect good 26008cd42b590dc71ee9c1ca667a218542aab342 # good: [92430b8fc8aac3b4ab92e9ca8a09d83c4788c609] Merge https://gitlab.denx.de/u-boot/custodians/u-boot-socfpga git bisect good 92430b8fc8aac3b4ab92e9ca8a09d83c4788c609 # bad: [2d64a0f7e952f54375702fb2b854461e402ded9d] Merge branch '2019-07-29-master-imports' git bisect bad 2d64a0f7e952f54375702fb2b854461e402ded9d # bad: [49b10cb4926285b856b207c1f5bb40c75487f08b] gpio: fixes for gpio-hog support git bisect bad 49b10cb4926285b856b207c1f5bb40c75487f08b # bad: [e35d533c79c75e9ad68a2e78da52c8ce58e6ce0d] tools/logos: remove black background of U-Boot logo git bisect bad e35d533c79c75e9ad68a2e78da52c8ce58e6ce0d # bad: [b1307f884a913f52a201491053b6d221c4204f60] fit: Support compression for non-kernel components (e.g. FDT) git bisect bad b1307f884a913f52a201491053b6d221c4204f60 # good: [2090854cd2f228bab546f2718ccdbe1664830d3c] common: Move bootm_decomp_image() to image.c (as image_decomp()) git bisect good 2090854cd2f228bab546f2718ccdbe1664830d3c # first bad commit: [b1307f884a913f52a201491053b6d221c4204f60] fit: Support compression for non-kernel components (e.g. FDT) hs@xmglap:u-boot [(kein Branch, binäre Suche begonnen bei master)] $
base was current master:
- bbaf56eda0 - Merge tag 'u-boot-amlogic-20190731' of
https://gitlab.denx.de/u-boot/custodians/u-boot-amlogic
I am currently not in my office to dig deeper into it ... may you have an idea?
Thanks!
bye, Heiko
[1] fit image source (created from yocto) /dts-v1/;
/ { description = "U-Boot fitImage for Poky (Yocto Project Reference Distro)/4.19.59/k30rf"; #address-cells = <1>;
images {[...] ramdisk@1 { description = "k30rf-image-initramfs"; data = /incbin/("/work/hs/tbot2go/k30rf/repo/k30rf/build_k30rf/tmp/deploy/images/k30rf/k30rf-image-initramfs-k30rf.cpio.gz"); type = "ramdisk"; arch = "arm"; os = "linux"; compression = "gzip";
Back last year, when I started what Julius succeeded with this pach, we came to the conclusion that this "compression = " tag in the FIT image wasn't about the actual content but about what U-Boot should do with the content. I.e. in your case it means "U-Boot, please uncompress this!", while the old code just did not care about compression and left uncompression to the Kernel.
To retain this behaviour, it should be enough to delete the "compression = " line from your its. That way, U-Boot doesn't uncompress the ramdisk and the Kernel should do it like always.
Regards, Simon
Hello Simon,
Am 01.08.2019 um 14:33 schrieb Simon Goldschmidt:
Am 01.08.2019 um 10:06 schrieb Heiko Schocher:
Hello Julius,
since your commit (just gone into mainline):
b1307f884a91 ("fit: Support compression for non-kernel components (e.g. FDT)")
I see on an imx6ull based board when booting (FIT image with kernel, dtb and initramdisk), fit image source created from yocto, see [1]:
## Loading kernel from FIT Image at 84000000 ... Using 'conf@imx6ull-k30rf.dtb' configuration Trying 'kernel@1' kernel subimage Description: Linux kernel Type: Kernel Image Compression: uncompressed Data Start: 0x840000e8 Data Size: 5154568 Bytes = 4.9 MiB Architecture: ARM OS: Linux Load Address: 0x82000000 Entry Point: 0x82000000 Hash algo: sha1 Hash value: a3ba8a09b197c0d1acab834c69c5aea677d4a092 Verifying Hash Integrity ... sha1+ OK ## Loading ramdisk from FIT Image at 84000000 ... Using 'conf@imx6ull-k30rf.dtb' configuration Trying 'ramdisk@1' ramdisk subimage Description: k30rf-image-initramfs Type: RAMDisk Image Compression: gzip compressed Data Start: 0x844f0cf8 Data Size: 6769496 Bytes = 6.5 MiB Architecture: ARM OS: Linux Load Address: unavailable Entry Point: unavailable Hash algo: sha1 Hash value: 507d8b20b18f9bca0e9178c48f634df5da720f65 Verifying Hash Integrity ... sha1+ OK Uncompressing RAMDisk Image Error: Bad gzipped data Error decompressing ramdisk Ramdisk image is corrupt or invalid =>
I found your commit with "git bisect":
hs@xmglap:u-boot [(kein Branch, binäre Suche begonnen bei master)] $ git bisect good b1307f884a913f52a201491053b6d221c4204f60 is the first bad commit commit b1307f884a913f52a201491053b6d221c4204f60 Author: Julius Werner jwerner@chromium.org Date: Wed Jul 24 19:37:55 2019 -0700
fit: Support compression for non-kernel components (e.g. FDT)
This patch adds support for compressing non-kernel image nodes in a FIT image (kernel nodes could already be compressed previously). This can reduce the size of FIT images and therefore improve boot times (especially when an image bundles many different kernel FDTs). The images will automatically be decompressed on load.
This patch does not support extracting compatible strings from compressed FDTs, so it's not very helpful in conjunction with CONFIG_FIT_BEST_MATCH yet, but it can already be used in environments that select the configuration to load explicitly.
Signed-off-by: Julius Werner jwerner@chromium.org Reviewed-by: Simon Glass sjg@chromium.org Reviewed-by: Simon Goldschmidt simon.k.r.goldschmidt@gmail.com
:040000 040000 6280a41241c1c25a28845fa526c5c958e448e347 e3f8ed43853541f868676f077924e7ab242d4da6 M common :040000 040000 406b0c50c03633bad9d4a576d5662f4fa483fc6f 601624cee600c1c06640858519d93188cc312544 M test hs@xmglap:u-boot [(kein Branch, binäre Suche begonnen bei master)] $ git bisect log git bisect start # bad: [bbaf56eda0e63d6d28fbccae0f112ef88203eb4d] Merge tag 'u-boot-amlogic-20190731' of https://gitlab.denx.de/u-boot/custodians/u-boot-amlogic git bisect bad bbaf56eda0e63d6d28fbccae0f112ef88203eb4d # good: [a9aa4c5700c68c070d63a391b51ea8d341b6e8a6] Merge branch '2019-07-24-master-imports' git bisect good a9aa4c5700c68c070d63a391b51ea8d341b6e8a6 # good: [75551c8bfc9545e31ec2ce238cac3857904007b8] Merge branch '2019-07-26-ti-imports' git bisect good 75551c8bfc9545e31ec2ce238cac3857904007b8 # bad: [333755ef7b6f824366eed37ae068c20a4f25a123] Merge branch '2019-07-29-ti-imports' git bisect bad 333755ef7b6f824366eed37ae068c20a4f25a123 # good: [26008cd42b590dc71ee9c1ca667a218542aab342] rockchip: rv1108: Migrate to use common board file git bisect good 26008cd42b590dc71ee9c1ca667a218542aab342 # good: [92430b8fc8aac3b4ab92e9ca8a09d83c4788c609] Merge https://gitlab.denx.de/u-boot/custodians/u-boot-socfpga git bisect good 92430b8fc8aac3b4ab92e9ca8a09d83c4788c609 # bad: [2d64a0f7e952f54375702fb2b854461e402ded9d] Merge branch '2019-07-29-master-imports' git bisect bad 2d64a0f7e952f54375702fb2b854461e402ded9d # bad: [49b10cb4926285b856b207c1f5bb40c75487f08b] gpio: fixes for gpio-hog support git bisect bad 49b10cb4926285b856b207c1f5bb40c75487f08b # bad: [e35d533c79c75e9ad68a2e78da52c8ce58e6ce0d] tools/logos: remove black background of U-Boot logo git bisect bad e35d533c79c75e9ad68a2e78da52c8ce58e6ce0d # bad: [b1307f884a913f52a201491053b6d221c4204f60] fit: Support compression for non-kernel components (e.g. FDT) git bisect bad b1307f884a913f52a201491053b6d221c4204f60 # good: [2090854cd2f228bab546f2718ccdbe1664830d3c] common: Move bootm_decomp_image() to image.c (as image_decomp()) git bisect good 2090854cd2f228bab546f2718ccdbe1664830d3c # first bad commit: [b1307f884a913f52a201491053b6d221c4204f60] fit: Support compression for non-kernel components (e.g. FDT) hs@xmglap:u-boot [(kein Branch, binäre Suche begonnen bei master)] $
base was current master: * bbaf56eda0 - Merge tag 'u-boot-amlogic-20190731' of https://gitlab.denx.de/u-boot/custodians/u-boot-amlogic
I am currently not in my office to dig deeper into it ... may you have an idea?
Thanks!
bye, Heiko
[1] fit image source (created from yocto) /dts-v1/;
/ { description = "U-Boot fitImage for Poky (Yocto Project Reference Distro)/4.19.59/k30rf"; #address-cells = <1>;
images { [...] ramdisk@1 { description = "k30rf-image-initramfs"; data = /incbin/("/work/hs/tbot2go/k30rf/repo/k30rf/build_k30rf/tmp/deploy/images/k30rf/k30rf-image-initramfs-k30rf.cpio.gz");
type = "ramdisk"; arch = "arm"; os = "linux"; compression = "gzip";
Back last year, when I started what Julius succeeded with this pach, we came to the conclusion that this "compression = " tag in the FIT image wasn't about the actual content but about what U-Boot should do with the content. I.e. in your case it means "U-Boot, please uncompress this!", while the old code just did not care about compression and left uncompression to the Kernel.
Understand.
To retain this behaviour, it should be enough to delete the "compression = " line from your its. That way, U-Boot doesn't uncompress the ramdisk and the Kernel should do it like always.
Ah ... okay ... Hmm.. the its file gets created from yocto recipe:
https://git.yoctoproject.org/cgit.cgi/poky/plain/meta/classes/kernel-fitimag...
I never checked :-(
So this commit breaks now yocto builded FIT images (for boards which use FIT image with an compressed ramdisk)
Added Marek, as he introduced "Add basic fitImage support" in yocto.
Hmm... thinking while writting ... this breaks also boards which only do an U-Boot update, and want to boot "old" FIT images from a storage device... Do we really want this?
Your arguments are correct, and the commit also, but we perhaps break some boards, so U-Boot should be able to handle the old (wrong) way?
The "compression = " line in the generated its file is added in kernel-fitimage.class from commit:
commit 4aa6644f92975ded908ef99cf313466e0845e071 Author: Rick Altherr raltherr@google.com Date: Fri Jan 20 11:28:53 2017 -0800
kernel-fitimage: Use compressed ramdisks in FIT images if available
kernel-fitimage:fitimage_assemble() was calling copy_initramfs from kernel.bbclass which decompresses the initramfs cpio. Assume that if INITRAMFS_FSTYPES includes a compressed cpio, that is what it desired in the FIT image.
(From OE-Core rev: 842ad404b36e00c89f615a3f7db4a2d30062effa)
Signed-off-by: Rick Altherr raltherr@google.com Signed-off-by: Ross Burton ross.burton@intel.com Signed-off-by: Richard Purdie richard.purdie@linuxfoundation.org
So more than 2 years in, so for sure boards out, which we break now ...
Added Rick, Ross and Richard to cc too, may they have an idea, what we should do.
I think we should do:
- U-Boot Kconfig option: to set U-Boot into old (wrong) way or at least, ignore compressed line in ramdisk section. Last idea seems the best option for me, may we do this always?
- correct yocto class, to not set "compressed =" line in its file.
Ideas?
bye, Heiko
On Thu, Aug 01, 2019 at 04:38:51PM +0200, Heiko Schocher wrote:
Hello Simon,
Am 01.08.2019 um 14:33 schrieb Simon Goldschmidt:
Am 01.08.2019 um 10:06 schrieb Heiko Schocher:
Hello Julius,
since your commit (just gone into mainline):
b1307f884a91 ("fit: Support compression for non-kernel components (e.g. FDT)")
I see on an imx6ull based board when booting (FIT image with kernel, dtb and initramdisk), fit image source created from yocto, see [1]:
## Loading kernel from FIT Image at 84000000 ... Using 'conf@imx6ull-k30rf.dtb' configuration Trying 'kernel@1' kernel subimage Description: Linux kernel Type: Kernel Image Compression: uncompressed Data Start: 0x840000e8 Data Size: 5154568 Bytes = 4.9 MiB Architecture: ARM OS: Linux Load Address: 0x82000000 Entry Point: 0x82000000 Hash algo: sha1 Hash value: a3ba8a09b197c0d1acab834c69c5aea677d4a092 Verifying Hash Integrity ... sha1+ OK ## Loading ramdisk from FIT Image at 84000000 ... Using 'conf@imx6ull-k30rf.dtb' configuration Trying 'ramdisk@1' ramdisk subimage Description: k30rf-image-initramfs Type: RAMDisk Image Compression: gzip compressed Data Start: 0x844f0cf8 Data Size: 6769496 Bytes = 6.5 MiB Architecture: ARM OS: Linux Load Address: unavailable Entry Point: unavailable Hash algo: sha1 Hash value: 507d8b20b18f9bca0e9178c48f634df5da720f65 Verifying Hash Integrity ... sha1+ OK Uncompressing RAMDisk Image Error: Bad gzipped data Error decompressing ramdisk Ramdisk image is corrupt or invalid =>
I found your commit with "git bisect":
hs@xmglap:u-boot [(kein Branch, binäre Suche begonnen bei master)] $ git bisect good b1307f884a913f52a201491053b6d221c4204f60 is the first bad commit commit b1307f884a913f52a201491053b6d221c4204f60 Author: Julius Werner jwerner@chromium.org Date: Wed Jul 24 19:37:55 2019 -0700
fit: Support compression for non-kernel components (e.g. FDT)
This patch adds support for compressing non-kernel image nodes in a FIT image (kernel nodes could already be compressed previously). This can reduce the size of FIT images and therefore improve boot times (especially when an image bundles many different kernel FDTs). The images will automatically be decompressed on load.
This patch does not support extracting compatible strings from compressed FDTs, so it's not very helpful in conjunction with CONFIG_FIT_BEST_MATCH yet, but it can already be used in environments that select the configuration to load explicitly.
Signed-off-by: Julius Werner jwerner@chromium.org Reviewed-by: Simon Glass sjg@chromium.org Reviewed-by: Simon Goldschmidt simon.k.r.goldschmidt@gmail.com
:040000 040000 6280a41241c1c25a28845fa526c5c958e448e347 e3f8ed43853541f868676f077924e7ab242d4da6 M common :040000 040000 406b0c50c03633bad9d4a576d5662f4fa483fc6f 601624cee600c1c06640858519d93188cc312544 M test hs@xmglap:u-boot [(kein Branch, binäre Suche begonnen bei master)] $ git bisect log git bisect start # bad: [bbaf56eda0e63d6d28fbccae0f112ef88203eb4d] Merge tag 'u-boot-amlogic-20190731' of https://gitlab.denx.de/u-boot/custodians/u-boot-amlogic git bisect bad bbaf56eda0e63d6d28fbccae0f112ef88203eb4d # good: [a9aa4c5700c68c070d63a391b51ea8d341b6e8a6] Merge branch '2019-07-24-master-imports' git bisect good a9aa4c5700c68c070d63a391b51ea8d341b6e8a6 # good: [75551c8bfc9545e31ec2ce238cac3857904007b8] Merge branch '2019-07-26-ti-imports' git bisect good 75551c8bfc9545e31ec2ce238cac3857904007b8 # bad: [333755ef7b6f824366eed37ae068c20a4f25a123] Merge branch '2019-07-29-ti-imports' git bisect bad 333755ef7b6f824366eed37ae068c20a4f25a123 # good: [26008cd42b590dc71ee9c1ca667a218542aab342] rockchip: rv1108: Migrate to use common board file git bisect good 26008cd42b590dc71ee9c1ca667a218542aab342 # good: [92430b8fc8aac3b4ab92e9ca8a09d83c4788c609] Merge https://gitlab.denx.de/u-boot/custodians/u-boot-socfpga git bisect good 92430b8fc8aac3b4ab92e9ca8a09d83c4788c609 # bad: [2d64a0f7e952f54375702fb2b854461e402ded9d] Merge branch '2019-07-29-master-imports' git bisect bad 2d64a0f7e952f54375702fb2b854461e402ded9d # bad: [49b10cb4926285b856b207c1f5bb40c75487f08b] gpio: fixes for gpio-hog support git bisect bad 49b10cb4926285b856b207c1f5bb40c75487f08b # bad: [e35d533c79c75e9ad68a2e78da52c8ce58e6ce0d] tools/logos: remove black background of U-Boot logo git bisect bad e35d533c79c75e9ad68a2e78da52c8ce58e6ce0d # bad: [b1307f884a913f52a201491053b6d221c4204f60] fit: Support compression for non-kernel components (e.g. FDT) git bisect bad b1307f884a913f52a201491053b6d221c4204f60 # good: [2090854cd2f228bab546f2718ccdbe1664830d3c] common: Move bootm_decomp_image() to image.c (as image_decomp()) git bisect good 2090854cd2f228bab546f2718ccdbe1664830d3c # first bad commit: [b1307f884a913f52a201491053b6d221c4204f60] fit: Support compression for non-kernel components (e.g. FDT) hs@xmglap:u-boot [(kein Branch, binäre Suche begonnen bei master)] $
base was current master: * bbaf56eda0 - Merge tag 'u-boot-amlogic-20190731' of https://gitlab.denx.de/u-boot/custodians/u-boot-amlogic
I am currently not in my office to dig deeper into it ... may you have an idea?
Thanks!
bye, Heiko
[1] fit image source (created from yocto) /dts-v1/;
/ { description = "U-Boot fitImage for Poky (Yocto Project Reference Distro)/4.19.59/k30rf"; #address-cells = <1>;
images { [...] ramdisk@1 { description = "k30rf-image-initramfs"; data = /incbin/("/work/hs/tbot2go/k30rf/repo/k30rf/build_k30rf/tmp/deploy/images/k30rf/k30rf-image-initramfs-k30rf.cpio.gz");
type = "ramdisk"; arch = "arm"; os = "linux"; compression = "gzip";
Back last year, when I started what Julius succeeded with this pach, we came to the conclusion that this "compression = " tag in the FIT image wasn't about the actual content but about what U-Boot should do with the content. I.e. in your case it means "U-Boot, please uncompress this!", while the old code just did not care about compression and left uncompression to the Kernel.
Understand.
To retain this behaviour, it should be enough to delete the "compression = " line from your its. That way, U-Boot doesn't uncompress the ramdisk and the Kernel should do it like always.
Ah ... okay ... Hmm.. the its file gets created from yocto recipe:
https://git.yoctoproject.org/cgit.cgi/poky/plain/meta/classes/kernel-fitimag...
I never checked :-(
So this commit breaks now yocto builded FIT images (for boards which use FIT image with an compressed ramdisk)
Added Marek, as he introduced "Add basic fitImage support" in yocto.
Hmm... thinking while writting ... this breaks also boards which only do an U-Boot update, and want to boot "old" FIT images from a storage device... Do we really want this?
Your arguments are correct, and the commit also, but we perhaps break some boards, so U-Boot should be able to handle the old (wrong) way?
The "compression = " line in the generated its file is added in kernel-fitimage.class from commit:
commit 4aa6644f92975ded908ef99cf313466e0845e071 Author: Rick Altherr raltherr@google.com Date: Fri Jan 20 11:28:53 2017 -0800
kernel-fitimage: Use compressed ramdisks in FIT images if available kernel-fitimage:fitimage_assemble() was calling copy_initramfs from kernel.bbclass which decompresses the initramfs cpio. Assume that if INITRAMFS_FSTYPES includes a compressed cpio, that is what it desired in the FIT image. (From OE-Core rev: 842ad404b36e00c89f615a3f7db4a2d30062effa) Signed-off-by: Rick Altherr <raltherr@google.com> Signed-off-by: Ross Burton <ross.burton@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>So more than 2 years in, so for sure boards out, which we break now ...
Added Rick, Ross and Richard to cc too, may they have an idea, what we should do.
I think we should do:
U-Boot Kconfig option: to set U-Boot into old (wrong) way or at least, ignore compressed line in ramdisk section. Last idea seems the best option for me, may we do this always?
correct yocto class, to not set "compressed =" line in its file.
First, we have a problem as we need to support what's out in the world.
Second, what is intended with what's out in the world? Indeed before we didn't do anything with the compressed data. But now what, we're saying it's compressed, but not providing enough details to U-Boot to uncompress it?
First, we have a problem as we need to support what's out in the world.
How about I write a patch to change the decompression call in fit_image_load() to fall back to a memcpy() if the decompression fails (and print a big warning that the FIT image is wrong and should be updated)? That should keep those old installations working.
Second, what is intended with what's out in the world? Indeed before we didn't do anything with the compressed data. But now what, we're saying it's compressed, but not providing enough details to U-Boot to uncompress it?
Hmm... I'm not sure why it wouldn't just decompress correctly, actually. Is there anything special with how gzipped data is formatted for the kernel decompressor? Or is there any limitation in the U-Boot gzip implementation?
Am 01.08.2019 um 20:24 schrieb Julius Werner:
First, we have a problem as we need to support what's out in the world.
How about I write a patch to change the decompression call in fit_image_load() to fall back to a memcpy() if the decompression fails (and print a big warning that the FIT image is wrong and should be updated)? That should keep those old installations working.
Well, I'm sure you'd find enough FIT images where the uncompression might succeed but the target space for decompression is not big enough (see bootm_size), would such an error be fixed as well?
Regards, Simon
Second, what is intended with what's out in the world? Indeed before we didn't do anything with the compressed data. But now what, we're saying it's compressed, but not providing enough details to U-Boot to uncompress it?
Hmm... I'm not sure why it wouldn't just decompress correctly, actually. Is there anything special with how gzipped data is formatted for the kernel decompressor? Or is there any limitation in the U-Boot gzip implementation?
Well, I'm sure you'd find enough FIT images where the uncompression might succeed but the target space for decompression is not big enough (see bootm_size), would such an error be fixed as well?
I think so? Assuming all the compression algorithms check for an output buffer overrun (looks like they do), that should also just lead to a compression error and go into the fallback path.
Am 01.08.2019 um 21:01 schrieb Julius Werner:
Well, I'm sure you'd find enough FIT images where the uncompression might succeed but the target space for decompression is not big enough (see bootm_size), would such an error be fixed as well?
I think so? Assuming all the compression algorithms check for an output buffer overrun (looks like they do), that should also just lead to a compression error and go into the fallback path.
Then I think that would be the best way to start and see if it's good enough for backwards compatibility.
Regards, Simon
Hello Julius,
Am 01.08.2019 um 20:24 schrieb Julius Werner:
First, we have a problem as we need to support what's out in the world.
How about I write a patch to change the decompression call in fit_image_load() to fall back to a memcpy() if the decompression fails (and print a big warning that the FIT image is wrong and should be updated)? That should keep those old installations working.
Hmm.. what happens if decompression does not fail ? (I wonder, why I get this Error, but I go into vacation, so I have no chance to dig deeper into it soon ...)
And why should we decompress, if we "know" we do not have to (in ramdisk case)? This cost only boottime ...
For the fast track, I prefer to ignore the compression property in ramdisk node ... (may configurable through a Kconfig option with default to ignore)
Second, what is intended with what's out in the world? Indeed before we didn't do anything with the compressed data. But now what, we're saying it's compressed, but not providing enough details to U-Boot to uncompress it?
Hmm... I'm not sure why it wouldn't just decompress correctly, actually. Is there anything special with how gzipped data is formatted for the kernel decompressor? Or is there any limitation in the U-Boot gzip implementation?
I do not know.
bye, Heiko
Hi Heiko,
On Fri, Aug 2, 2019 at 6:08 AM Heiko Schocher hs@denx.de wrote:
Hello Julius,
Am 01.08.2019 um 20:24 schrieb Julius Werner:
First, we have a problem as we need to support what's out in the world.
How about I write a patch to change the decompression call in fit_image_load() to fall back to a memcpy() if the decompression fails (and print a big warning that the FIT image is wrong and should be updated)? That should keep those old installations working.
Hmm.. what happens if decompression does not fail ? (I wonder, why I get this Error, but I go into vacation, so I have no chance to dig deeper into it soon ...)
And why should we decompress, if we "know" we do not have to (in ramdisk case)? This cost only boottime ...
It's correct we don't need to decompress since the kernel can decompress the ramdisk. However, the FIT image is wrong (given our updated definition of the "compressed" field) and it's only decompressed on "legacy" images.
For the fast track, I prefer to ignore the compression property in ramdisk node ... (may configurable through a Kconfig option with default to ignore)
I'm not too fond of adding such extra cases, but you're right that decompressing a ramdisk is probably never needed...
Regards, Simon
Second, what is intended with what's out in the world? Indeed before we didn't do anything with the compressed data. But now what, we're saying it's compressed, but not providing enough details to U-Boot to uncompress it?
Hmm... I'm not sure why it wouldn't just decompress correctly, actually. Is there anything special with how gzipped data is formatted for the kernel decompressor? Or is there any limitation in the U-Boot gzip implementation?
I do not know.
bye, Heiko -- DENX Software Engineering GmbH, Managing Director: Wolfgang Denk HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: +49-8142-66989-52 Fax: +49-8142-66989-80 Email: hs@denx.de
On Fri, Aug 02, 2019 at 08:00:22AM +0200, Simon Goldschmidt wrote:
Hi Heiko,
On Fri, Aug 2, 2019 at 6:08 AM Heiko Schocher hs@denx.de wrote:
Hello Julius,
Am 01.08.2019 um 20:24 schrieb Julius Werner:
First, we have a problem as we need to support what's out in the world.
How about I write a patch to change the decompression call in fit_image_load() to fall back to a memcpy() if the decompression fails (and print a big warning that the FIT image is wrong and should be updated)? That should keep those old installations working.
Hmm.. what happens if decompression does not fail ? (I wonder, why I get this Error, but I go into vacation, so I have no chance to dig deeper into it soon ...)
And why should we decompress, if we "know" we do not have to (in ramdisk case)? This cost only boottime ...
It's correct we don't need to decompress since the kernel can decompress the ramdisk. However, the FIT image is wrong (given our updated definition of the "compressed" field) and it's only decompressed on "legacy" images.
For the fast track, I prefer to ignore the compression property in ramdisk node ... (may configurable through a Kconfig option with default to ignore)
I'm not too fond of adding such extra cases, but you're right that decompressing a ramdisk is probably never needed...
I think we have a case where intentions differed historically and now we have to deal with it. Back in the earlier thread this year where we talked about this exact case, it was OK to start decompressing ramdisks as that was the intention way back when, but it was silently broken forever ago. After it was broken is when FIT adoption finally took off and we have the real life case of "set compression to the factual value, but it's not used by U-Boot!" has now become "set compression to the factual value and U-Boot decompresses it".
Maybe the best choice is to rework things so that we only check compression on (and decompress) device trees and when we print ramdisk information we note that it's not being decompressed?
On Fri, Aug 2, 2019 at 2:36 PM Tom Rini trini@konsulko.com wrote:
On Fri, Aug 02, 2019 at 08:00:22AM +0200, Simon Goldschmidt wrote:
Hi Heiko,
On Fri, Aug 2, 2019 at 6:08 AM Heiko Schocher hs@denx.de wrote:
Hello Julius,
Am 01.08.2019 um 20:24 schrieb Julius Werner:
First, we have a problem as we need to support what's out in the world.
How about I write a patch to change the decompression call in fit_image_load() to fall back to a memcpy() if the decompression fails (and print a big warning that the FIT image is wrong and should be updated)? That should keep those old installations working.
Hmm.. what happens if decompression does not fail ? (I wonder, why I get this Error, but I go into vacation, so I have no chance to dig deeper into it soon ...)
And why should we decompress, if we "know" we do not have to (in ramdisk case)? This cost only boottime ...
It's correct we don't need to decompress since the kernel can decompress the ramdisk. However, the FIT image is wrong (given our updated definition of the "compressed" field) and it's only decompressed on "legacy" images.
For the fast track, I prefer to ignore the compression property in ramdisk node ... (may configurable through a Kconfig option with default to ignore)
I'm not too fond of adding such extra cases, but you're right that decompressing a ramdisk is probably never needed...
I think we have a case where intentions differed historically and now we have to deal with it. Back in the earlier thread this year where we talked about this exact case, it was OK to start decompressing ramdisks as that was the intention way back when, but it was silently broken forever ago. After it was broken is when FIT adoption finally took off and we have the real life case of "set compression to the factual value, but it's not used by U-Boot!" has now become "set compression to the factual value and U-Boot decompresses it".
Maybe the best choice is to rework things so that we only check compression on (and decompress) device trees and when we print ramdisk information we note that it's not being decompressed?
While I can understand that point of view, please note that my intention is not having a compressed devicetree (where we save some KByte) but having a compressed FPGA image (where I save about 2 MByte!).
Given that, please don't start a positive list (e.g. only decompress devicetree and Kernel). But if we actually need to do this differently per type, just implement Heiko's suggestion and don't decompress ramdisks.
That way, we could show an error when seeing this setting for ramdisks and probably remove this special case handling in some years...?
Regards, Simon
On Fri, Aug 02, 2019 at 02:50:40PM +0200, Simon Goldschmidt wrote:
On Fri, Aug 2, 2019 at 2:36 PM Tom Rini trini@konsulko.com wrote:
On Fri, Aug 02, 2019 at 08:00:22AM +0200, Simon Goldschmidt wrote:
Hi Heiko,
On Fri, Aug 2, 2019 at 6:08 AM Heiko Schocher hs@denx.de wrote:
Hello Julius,
Am 01.08.2019 um 20:24 schrieb Julius Werner:
First, we have a problem as we need to support what's out in the world.
How about I write a patch to change the decompression call in fit_image_load() to fall back to a memcpy() if the decompression fails (and print a big warning that the FIT image is wrong and should be updated)? That should keep those old installations working.
Hmm.. what happens if decompression does not fail ? (I wonder, why I get this Error, but I go into vacation, so I have no chance to dig deeper into it soon ...)
And why should we decompress, if we "know" we do not have to (in ramdisk case)? This cost only boottime ...
It's correct we don't need to decompress since the kernel can decompress the ramdisk. However, the FIT image is wrong (given our updated definition of the "compressed" field) and it's only decompressed on "legacy" images.
For the fast track, I prefer to ignore the compression property in ramdisk node ... (may configurable through a Kconfig option with default to ignore)
I'm not too fond of adding such extra cases, but you're right that decompressing a ramdisk is probably never needed...
I think we have a case where intentions differed historically and now we have to deal with it. Back in the earlier thread this year where we talked about this exact case, it was OK to start decompressing ramdisks as that was the intention way back when, but it was silently broken forever ago. After it was broken is when FIT adoption finally took off and we have the real life case of "set compression to the factual value, but it's not used by U-Boot!" has now become "set compression to the factual value and U-Boot decompresses it".
Maybe the best choice is to rework things so that we only check compression on (and decompress) device trees and when we print ramdisk information we note that it's not being decompressed?
While I can understand that point of view, please note that my intention is not having a compressed devicetree (where we save some KByte) but having a compressed FPGA image (where I save about 2 MByte!).
Given that, please don't start a positive list (e.g. only decompress devicetree and Kernel). But if we actually need to do this differently per type, just implement Heiko's suggestion and don't decompress ramdisks.
That way, we could show an error when seeing this setting for ramdisks and probably remove this special case handling in some years...?
OK. So lets special case ramdisk to just note that we see it's compressed ("compression: %s, ignored" ?) and then we'll find out if there's really a bunch of cases where people assumed it was decompressed by us or not.
participants (4)
-
Heiko Schocher -
Julius Werner -
Simon Goldschmidt -
Tom Rini