Skip to contentSkip to conversation

squashfs-4.0: update squashfs-tools to include upstream LZMA inclusion.

PATCH In review Leon Woestenberg 2 participants 5 messages Nov 18, 2009

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 squashfs-4.0: update squashfs-tools to include upstream LZMA inclusion.

In OpenEmbedded we used squashfs 4.0 with external patch for LZMA compression support. In the meantime, squashfs has mainlined LZMA support in its CVS version.

This patches switches to the upstream squashfs-tools and modifies bitbake.conf to match the slightly different arguments to mksquashfs.

---
 conf/bitbake.conf                                  |    2 +-
 conf/distro/include/sane-srcdates.inc              |    1 +
 .../squashfs-tools/squashfs-tools-native_4.0.bb    |    7 +--
 recipes/squashfs-tools/squashfs-tools_4.0.bb       |   42 +++++++++++++++-----
 4 files changed, 37 insertions(+), 15 deletions(-)
conf/bitbake.conf +1 -1
index 96e7799..bb99d3c 100644
@@ -345,7 +345,7 @@ IMAGE_CMD_ext2.gz = "install -d ${DEPLOY_DIR_IMAGE}/tmp.gz ; genext2fs -b ${ROOT
 IMAGE_CMD_ext3 = "genext2fs -b ${ROOTFS_SIZE} -d ${IMAGE_ROOTFS} ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.ext3 ${EXTRA_IMAGECMD}; tune2fs -j ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.ext3"
 IMAGE_CMD_ext3.gz = "install -d ${DEPLOY_DIR_IMAGE}/tmp.gz ; genext2fs -b ${ROOTFS_SIZE} -d ${IMAGE_ROOTFS} ${DEPLOY_DIR_IMAGE}/tmp.gz/${IMAGE_NAME}.rootfs.ext3 ${EXTRA_IMAGECMD}; tune2fs -j ${DEPLOY_DIR_IMAGE}/tmp.gz/${IMAGE_NAME}.rootfs.ext3; gzip -f -9 ${DEPLOY_DIR_IMAGE}/tmp.gz/${IMAGE_NAME}.rootfs.ext3; mv ${DEPLOY_DIR_IMAGE}/tmp.gz/${IMAGE_NAME}.rootfs.ext3.gz ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.ext3.gz"
 IMAGE_CMD_squashfs = "mksquashfs ${IMAGE_ROOTFS} ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.squashfs ${EXTRA_IMAGECMD} -noappend"
-IMAGE_CMD_squashfs-lzma = "mksquashfs ${IMAGE_ROOTFS} ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.squashfs-lzma ${EXTRA_IMAGECMD} -noappend -lzma"
+IMAGE_CMD_squashfs-lzma = "mksquashfs ${IMAGE_ROOTFS} ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.squashfs-lzma ${EXTRA_IMAGECMD} -noappend -comp lzma"
 IMAGE_CMD_tar = "cd ${IMAGE_ROOTFS} && tar -cvf ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.tar ."
 IMAGE_CMD_tar.gz = "cd ${IMAGE_ROOTFS} && tar -zcvf ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.tar.gz ."
 IMAGE_CMD_tar.bz2 = "cd ${IMAGE_ROOTFS} && tar -jcvf ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.tar.bz2 ."
conf/distro/include/sane-srcdates.inc +1 -0
index 422145c..4ea318f 100644
@@ -41,6 +41,7 @@ SRCDATE_rdesktop ?= "20080917"
 SRCDATE_roadster ?= "20060814"
 SRCDATE_rosetta ?= "20090514"
 SRCDATE_sctzap ?= "20060814"
+SRCDATE_squashfs ?= "20091110"
 SRCDATE_tslib ?= "20051101"
 SRCDATE_waimea ?= "20060814"
 SRCDATE_xcompmgr ?= "20060814"
recipes/squashfs-tools/squashfs-tools-native_4.0.bb +3 -4
index 8fc80a6..cc01604 100644
@@ -1,13 +1,12 @@
-require squashfs-tools_${PV}.bb
+require squashfs-tools_4.0.bb
 
 inherit native
 
-DEPENDS = "zlib-native lzma-native "
-
-PR = "${INC_PR}.1"
+DEPENDS = "zlib-native"
 
 PACKAGES = ""
 
 do_stage () {
 	install -m 0755 mksquashfs ${STAGING_BINDIR}/
 }
+
recipes/squashfs-tools/squashfs-tools_4.0.bb +32 -10

46 lines not on this page · raw message

Signature · 1 line
1.6.0.4
#1 lore raw

Hello Leon,

Quoted 6 lines · Leon Woestenberg

In OpenEmbedded we used squashfs 4.0 with external patch for LZMA compression support. In the meantime, squashfs has mainlined LZMA support in its CVS version.

This patches switches to the upstream squashfs-tools and modifies bitbake.conf to match the slightly different arguments to mksquashfs.

I agree with the move to mainline version however your recipe changes might need some improvement. You don't use the lzma recipe available in OE and instead grab it again ... IMO this is wrong since it duplicates the lzma source and also make more difficult to us to "share" fixes.

Besides that I think it is ok. Can you fix it and send a new patch for review?

Cheers,

Signature · 3 lines
Otavio Salvador                  O.S. Systems
E-mail: [email protected]  http://www.ossystems.com.br
Mobile: +55 53 9981-7854         http://projetos.ossystems.com.br
#2 lore raw

Hello Otavio,

Quoted 6 lines · Otavio Salvador
Quoted 4 lines · Leon Woestenberg

In OpenEmbedded we used squashfs 4.0 with external patch for LZMA compression support. In the meantime, squashfs has mainlined LZMA support in its CVS version.

Quoted 9 lines
Quoted 2 lines

This patches switches to the upstream squashfs-tools and modifies bitbake.conf to match the slightly different arguments to mksquashfs.

I agree with the move to mainline version however your recipe changes might need some improvement. You don't use the lzma recipe available in OE and instead grab it again ... IMO this is wrong since it duplicates the lzma source and also make more difficult to us to "share" fixes.

Yes, I noticed this.

However, this is *exactl*y how the squashfs author designed his integration of LZMA; simply referring to a specific extracted archive of the LZMA SDK.

(consider this is a lazy man's way of static linking)

Quoted 2 lines

Besides that  I think it is ok. Can you fix it and send a new patch for review?

I will not fix it. Instead I will contact the squashfs author to ask if he want to consider the neat approach (if of course LZMA provides this).

In the meantime, should I apply the patch as-is?

Regards,

Signature · 1 line
Leon
#3 lore raw

Hello,

Quoted 6 lines · Leon Woestenberg
Quoted 6 lines

I agree with the move to mainline version however your recipe changes might need some improvement. You don't use the lzma recipe available in OE and instead grab it again ... IMO this is wrong since it duplicates the lzma source and also make more difficult to us to "share" fixes.

Quoted 7 lines

Yes, I noticed this.

However, this is *exactl*y how the squashfs author designed his integration of LZMA; simply referring to a specific extracted archive of the LZMA SDK.

(consider this is a lazy man's way of static linking)

Anyway I think it is worth fixing it.

Duplicating sources (and the needed fixes, in case LZMA requires one) is wrong if avoidable so I'd say it would be nice to fix it.

Quoted 5 lines
Quoted 2 lines

Besides that  I think it is ok. Can you fix it and send a new patch for review?

I will not fix it. Instead I will contact the squashfs author to ask if he want to consider the neat approach (if of course LZMA provides this).

Look above.

Quoted 1 line

In the meantime, should I apply the patch as-is?

You already did but I'd expect some time to reply before you pushing it.

:-(

Signature · 3 lines
Otavio Salvador                  O.S. Systems
E-mail: [email protected]  http://www.ossystems.com.br
Mobile: +55 53 9981-7854         http://projetos.ossystems.com.br
#4 lore raw

Hello Otavio,

Quoted 10 lines · Otavio Salvador
Quoted 8 lines · Leon Woestenberg
Quoted 6 lines

I agree with the move to mainline version however your recipe changes might need some improvement. You don't use the lzma recipe available in OE and instead grab it again ... IMO this is wrong since it duplicates the lzma source and also make more difficult to us to "share" fixes.

Yes, I noticed this.

Quoted 8 lines
Quoted 5 lines

However, this is *exactl*y how the squashfs author designed his integration of LZMA; simply referring to a specific extracted archive of the LZMA SDK.

(consider this is a lazy man's way of static linking)

Anyway I think it is worth fixing it.

Yes, agreed, only if that approach would work (i.e. the needed functions are correctly exported into the static library) with the current code base.

Quoted 3 lines

Duplicating sources (and the needed fixes, in case LZMA requires one) is wrong if avoidable so I'd say it would be nice to fix it.

Duplicating effort working against the upstream package is wrong too. I would rather
go through the author to see this approach mainlined.

Quoted 3 lines
Quoted 1 line

In the meantime, should I apply the patch as-is?

You already did but I'd expect some time to reply before you pushing it.

Sorry I misunderstood your "good and thx" reply on IRC. The current lzma recipe has some problems (use of dos2unix, not building for target) that we need to fix first.

The currently committed recipe is pure metadata, builds the same way as the upstream package does, and is tested for -native on a target, by booting a squashfs-lzma compressed rootfs.
I propose to keep it as-is, then see if we can fix lzma, neaten the squashfs-tools build, and bring it upstream.

Regards,

Signature · 1 line
Leon