opkg: enable zstd support
PATCH 1/1 opkg: enable zstd support
This allows the use of zstd for opkg packages by using OPKGBUILDCMD:
OPKGBUILDCMD = "opkg-build -Z zstd"
--- meta/recipes-devtools/opkg/opkg_0.6.0.bb | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
meta/recipes-devtools/opkg/opkg_0.6.0.bb +2 -1
@@ -30,7 +30,7 @@ inherit autotools pkgconfig ptest target_localstatedir := "${localstatedir}" OPKGLIBDIR ??= "${target_localstatedir}/lib" -PACKAGECONFIG ??= "libsolv" +PACKAGECONFIG ??= "libsolv zstd" PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ gnupg gpgme libgpg-error,\
@@ -39,6 +39,7 @@ PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ PACKAGECONFIG[curl] = "--enable-curl,--disable-curl,curl" PACKAGECONFIG[ssl-curl] = "--enable-ssl-curl,--disable-ssl-curl,curl openssl" PACKAGECONFIG[sha256] = "--enable-sha256,--disable-sha256" +PACKAGECONFIG[zstd] = "--enable-zstd,--disable-zstd,zstd" PACKAGECONFIG[libsolv] = "--with-libsolv,--without-libsolv,libsolv" EXTRA_OECONF:class-native = "--localstatedir=/${@os.path.relpath('${localstatedir}', '${STAGING_DIR_NATIVE}')} --sysconfdir=/${@os.path.relpath('${sysconfdir}', '${STAGING_DIR_NATIVE}')}"
Signature · 1 line
2.36.1.vfs.0.0
ACK from me - apart from enabling zstd by default.
Quoted 14 lines · Etienne Cordonnier via lists.openembedded.org
This allows the use of zstd for opkg packages by using OPKGBUILDCMD:
OPKGBUILDCMD = "opkg-build -Z zstd"Signed-off-by Alex Feinman <[email protected]>Signed-off-by Etienne Cordonnier <[email protected]>--- meta/recipes-devtools/opkg/opkg_0.6.0.bb | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)meta/recipes-devtools/opkg/opkg_0.6.0.bb +0
-0
meta/recipes-devtools/opkg/opkg_0.6.0.bb inherit autotools pkgconfig ptest · line 33
Quoted 5 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb +1
-1@@ -30,7 +30,7 @@ inherit autotools pkgconfig ptest target_localstatedir := "${localstatedir}" OPKGLIBDIR ??= "${target_localstatedir}/lib" -PACKAGECONFIG ??= "libsolv" +PACKAGECONFIG ??= "libsolv zstd"
Building in zstd support by default is a little suspect to me.
Unless I'm mistaken, OE-core will only build xz-compressed IPKs by default. So zstd support would be unnecessary for a distro integrator who just uses upstream OE-core.
For distros which use zstd compression in their packages, I think it would be more appropriate to overwrite the opkg PACKAGECONFIG in a .bbappend.
Is there something I'm not considering here?
Quoted 20 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb +1
-1PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ gnupg gpgme libgpg-error,\@@ -39,6 +39,7 @@ PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ PACKAGECONFIG[curl] = "--enable-curl,--disable-curl,curl" PACKAGECONFIG[ssl-curl] = "--enable-ssl-curl,--disable-ssl-curl,curl openssl" PACKAGECONFIG[sha256] = "--enable-sha256,--disable-sha256" +PACKAGECONFIG[zstd] = "--enable-zstd,--disable-zstd,zstd" PACKAGECONFIG[libsolv] = "--with-libsolv,--without-libsolv,libsolv" EXTRA_OECONF:class-native = "--localstatedir=/${@os.path.relpath('${localstatedir}', '${STAGING_DIR_NATIVE}')} --sysconfdir=/${@os.path.relpath('${sysconfdir}', '${STAGING_DIR_NATIVE}')}" -=-=-=-=-=-=-=-=-=-=-=-Links: You receive all messages sent to this group.
View/Reply Online (#170575): https://urldefense.com/v3/__https://lists.openembedded.org/g/openembedded-core/message/170575__;!!FbZ0ZwI3Qg!pogpLkfpLwPy4zdzhTDEFkMT6eNuynfXzyaNvTq4OK74eYha04b285qGN5bk1t7aQmJAqnSBddI6-TD8FLOtzgtG_5PjNUviztKz$
Mute This Topic: https://urldefense.com/v3/__https://lists.openembedded.org/mt/93654146/3616788__;!!FbZ0ZwI3Qg!pogpLkfpLwPy4zdzhTDEFkMT6eNuynfXzyaNvTq4OK74eYha04b285qGN5bk1t7aQmJAqnSBddI6-TD8FLOtzgtG_5PjNVdoKO7n$
Group Owner: [email protected]
Unsubscribe: https://urldefense.com/v3/__https://lists.openembedded.org/g/openembedded-core/unsub__;!!FbZ0ZwI3Qg!pogpLkfpLwPy4zdzhTDEFkMT6eNuynfXzyaNvTq4OK74eYha04b285qGN5bk1t7aQmJAqnSBddI6-TD8FLOtzgtG_5PjNfNktoah$ [[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
Signature · 5 lines
Alex Stewart Software Engineer - NI Real-Time OS NI (National Instruments) [email protected]
Quoted 26 lines · Alex Stewart
ACK from me - apart from enabling zstd by default.
Quoted 14 lines · Etienne Cordonnier via lists.openembedded.org
This allows the use of zstd for opkg packages by using OPKGBUILDCMD:
OPKGBUILDCMD = "opkg-build -Z zstd"Signed-off-by Alex Feinman <[email protected]>Signed-off-by Etienne Cordonnier <[email protected]>--- meta/recipes-devtools/opkg/opkg_0.6.0.bb | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)meta/recipes-devtools/opkg/opkg_0.6.0.bb +0
-0meta/recipes-devtools/opkg/opkg_0.6.0.bb inherit autotools pkgconfig ptest · line 33
Quoted 5 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb +1
-1@@ -30,7 +30,7 @@ inherit autotools pkgconfig ptest target_localstatedir := "${localstatedir}" OPKGLIBDIR ??= "${target_localstatedir}/lib" -PACKAGECONFIG ??= "libsolv" +PACKAGECONFIG ??= "libsolv zstd"Building in zstd support by default is a little suspect to me.
Quoted 8 lines
Unless I'm mistaken, OE-core will only build xz-compressed IPKs by default. So zstd support would be unnecessary for a distro integrator who just uses upstream OE-core.
For distros which use zstd compression in their packages, I think it would be more appropriate to overwrite the opkg PACKAGECONFIG in a .bbappend.
This is perhaps fine. I do wonder if there is some performance comparison data between xz and zstd compressed ipks
with opkg, it might help users on making this choice and also if we should consider using
zstd by default at some point or not.
Quoted 33 lines
Is there something I'm not considering here?
Quoted 7 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb +0
-0PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ gnupg gpgme libgpg-error,\meta/recipes-devtools/opkg/opkg_0.6.0.bb PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ · line 45
Quoted 7 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb +1
-0@@ -39,6 +39,7 @@ PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ PACKAGECONFIG[curl] = "--enable-curl,--disable-curl,curl" PACKAGECONFIG[ssl-curl] = "--enable-ssl-curl,--disable-ssl-curl,curl openssl" PACKAGECONFIG[sha256] = "--enable-sha256,--disable-sha256" +PACKAGECONFIG[zstd] = "--enable-zstd,--disable-zstd,zstd" PACKAGECONFIG[libsolv] = "--with-libsolv,--without-libsolv,libsolv" EXTRA_OECONF:class-native = "--localstatedir=/${@os.path.relpath('${localstatedir}', '${STAGING_DIR_NATIVE}')} --sysconfdir=/${@os.path.relpath('${sysconfdir}', '${STAGING_DIR_NATIVE}')}"--
Alex Stewart
Software Engineer - NI Real-Time OS
NI (National Instruments)-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#170608): https://lists.openembedded.org/g/openembedded-core/message/170608
Mute This Topic: https://lists.openembedded.org/mt/93654146/1997914
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
A message that is not in the archive (sent off-list or not yet received): CANOoYsMVb9YpfwjTNHv4kSARM15_3T34CA-Ekr=t7P5F0XO4bA@mail.gmail.com
Quoted 113 lines · Alex Feinman
I do have some numbers. When I was selling this change internally, I did a comparison on our internal build.
Combined write IPK times (Σ t
do_package_write_ipk)xz 162m 35s
gz 52m 13s
zstd 33m 49s
Compression rate for zstd was closer to xz than to gz but not as good as xz. For systems that have to cache packages on the device with limited storage xz might be a better option, but for the bulk of projects zstd is the best choice
Additionally, zstd offers much faster decompression than xz so the rootfs build step that includes unpacking all of the ipks, takes 3m 58s with xz and 2m 44s with zstd.
One other thing of note - if your build includes debug packages, some may be quite large. E.g. one of our components produces a 2.2 GB debug package (uncompressed). On large files xz requires a disproportionally large amount of time resulting in 15 minutes needed to simply write ipk for the abovementioned packages, whereas zstd took about 45 sec. For frequent tasks like bitbaking a single package this translates in a lot of saved time.
Bottom line - I think making xz a default package compressor was not entirely thought through. gzip or zstd is what the default should be.
One final note: I could not find a reasonable explanation for why opkg-tools require code changes to support a different compressor. BSD tar and GNU tar both can easily accept compressors that they have no idea about (via -I option) because all of them provide a unified command line interface for use in pipes. If this were done similar to tar, we could have used any compressor we wanted, including the multithreaded versions (zstdmt)
Quoted 88 lines · Khem Raj
Quoted 7 lines · Alex Stewart
ACK from me - apart from enabling zstd by default.
Quoted 3 lines · Etienne Cordonnier via lists.openembedded.org
This allows the use of zstd for opkg packages by using OPKGBUILDCMD:
OPKGBUILDCMD = "opkg-build -Z zstd"Quoted 7 lines
Review
Quoted 7 lines
Signed-off-by Alex Feinman <[email protected]>Signed-off-by Etienne Cordonnier <[email protected]>--- meta/recipes-devtools/opkg/opkg_0.6.0.bb | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)(quoted diff) +0
-0b/meta/recipes-devtools/opkg/opkg_0.6.0.bb
Quoted 12 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb inherit autotools pkgconfig ptest · line 33
Quoted 9 lines
index
7b351e8123..e38d9d6f3f100644meta/recipes-devtools/opkg/opkg_0.6.0.bb +1
-1@@ -30,7 +30,7 @@ inherit autotools pkgconfig ptest target_localstatedir := "${localstatedir}" OPKGLIBDIR ??= "${target_localstatedir}/lib" -PACKAGECONFIG ??= "libsolv" +PACKAGECONFIG ??= "libsolv zstd"Building in zstd support by default is a little suspect to me.
Quoted 8 lines
Unless I'm mistaken, OE-core will only build xz-compressed IPKs by default. So zstd support would be unnecessary for a distro integrator who just uses upstream OE-core.
For distros which use zstd compression in their packages, I think it would be more appropriate to overwrite the opkg PACKAGECONFIG in a .bbappend.
This is perhaps fine. I do wonder if there is some performance comparison data between xz and zstd compressed ipks
with opkg, it might help users on making this choice and also if we should consider using
zstd by default at some point or not.Quoted 8 lines
Is there something I'm not considering here?
meta/recipes-devtools/opkg/opkg_0.6.0.bb PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ · line 40
Quoted 6 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb +0
-0PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ gnupg gpgme libgpg-error,\@@ -39,6 +39,7 @@ PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ PACKAGECONFIG[curl] = "--enable-curl,--disable-curl,curl" PACKAGECONFIG[ssl-curl] ="--enable-ssl-curl,--disable-ssl-curl,curl openssl"
Quoted 5 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\
Quoted 5 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb +1
-0PACKAGECONFIG[sha256] = "--enable-sha256,--disable-sha256" +PACKAGECONFIG[zstd] = "--enable-zstd,--disable-zstd,zstd" PACKAGECONFIG[libsolv] = "--with-libsolv,--without-libsolv,libsolv" EXTRA_OECONF:class-native ="--localstatedir=/${@os.path.relpath('${localstatedir}', '${STAGING_DIR_NATIVE}')} --sysconfdir=/${@os.path.relpath('${sysconfdir}', '${STAGING_DIR_NATIVE}')}"
Quoted 9 lines
Quoted 3 lines
--
Alex Stewart
Software Engineer - NI Real-Time OS
NI (National Instruments)Quoted 6 lines
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group. View/Reply Online (#170608):Quoted 1 line
Mute This Topic:
Quoted 2 lines
Group Owner: [email protected] Unsubscribe:
https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_unsub&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=PxAE_3yGzbSZ21-LHz_1r3gd2_STPoZcrYDeu9vd_GM&e=
[[email protected]]Quoted 2 lines
-=-=-=-=-=-=-=-=-=-=-=-
Thanks for sharing, really appreciate it. How close were compression rates in numbers ? I think some users might be sensitive to the archive sizes, if its something
with in acceptable range, I do agree, zstd would be a better default these days.
Quoted 99 lines · Alex Feinman
I do have some numbers. When I was selling this change internally, I did a comparison on our internal build.
Combined write IPK times (Σ t
do_package_write_ipk)xz 162m 35s
gz 52m 13s
zstd 33m 49s
Compression rate for zstd was closer to xz than to gz but not as good as xz. For systems that have to cache packages on the device with limited storage xz might be a better option, but for the bulk of projects zstd is the best choice
Additionally, zstd offers much faster decompression than xz so the rootfs build step that includes unpacking all of the ipks, takes 3m 58s with xz and 2m 44s with zstd.
One other thing of note - if your build includes debug packages, some may be quite large. E.g. one of our components produces a 2.2 GB debug package (uncompressed). On large files xz requires a disproportionally large amount of time resulting in 15 minutes needed to simply write ipk for the abovementioned packages, whereas zstd took about 45 sec. For frequent tasks like bitbaking a single package this translates in a lot of saved time.
Bottom line - I think making xz a default package compressor was not entirely thought through. gzip or zstd is what the default should be.
One final note: I could not find a reasonable explanation for why opkg-tools require code changes to support a different compressor. BSD tar and GNU tar both can easily accept compressors that they have no idea about (via -I option) because all of them provide a unified command line interface for use in pipes. If this were done similar to tar, we could have used any compressor we wanted, including the multithreaded versions (zstdmt)
Quoted 76 lines · Khem Raj
Quoted 26 lines · Alex Stewart
ACK from me - apart from enabling zstd by default.
Quoted 14 lines · Etienne Cordonnier via lists.openembedded.org
This allows the use of zstd for opkg packages by using OPKGBUILDCMD:
OPKGBUILDCMD = "opkg-build -Z zstd"Signed-off-by Alex Feinman <[email protected]>Signed-off-by Etienne Cordonnier <[email protected]>--- meta/recipes-devtools/opkg/opkg_0.6.0.bb | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)meta/recipes-devtools/opkg/opkg_0.6.0.bb +0
-0meta/recipes-devtools/opkg/opkg_0.6.0.bb inherit autotools pkgconfig ptest · line 33
Quoted 5 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb +1
-1@@ -30,7 +30,7 @@ inherit autotools pkgconfig ptest target_localstatedir := "${localstatedir}" OPKGLIBDIR ??= "${target_localstatedir}/lib" -PACKAGECONFIG ??= "libsolv" +PACKAGECONFIG ??= "libsolv zstd"Building in zstd support by default is a little suspect to me.
Quoted 8 lines
Unless I'm mistaken, OE-core will only build xz-compressed IPKs by default. So zstd support would be unnecessary for a distro integrator who just uses upstream OE-core.
For distros which use zstd compression in their packages, I think it would be more appropriate to overwrite the opkg PACKAGECONFIG in a .bbappend.
This is perhaps fine. I do wonder if there is some performance comparison data between xz and zstd compressed ipks
with opkg, it might help users on making this choice and also if we should consider using
zstd by default at some point or not.Quoted 33 lines
Is there something I'm not considering here?
Quoted 7 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb +0
-0PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ gnupg gpgme libgpg-error,\meta/recipes-devtools/opkg/opkg_0.6.0.bb PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ · line 45
Quoted 7 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb +1
-0@@ -39,6 +39,7 @@ PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ PACKAGECONFIG[curl] = "--enable-curl,--disable-curl,curl" PACKAGECONFIG[ssl-curl] = "--enable-ssl-curl,--disable-ssl-curl,curl openssl" PACKAGECONFIG[sha256] = "--enable-sha256,--disable-sha256" +PACKAGECONFIG[zstd] = "--enable-zstd,--disable-zstd,zstd" PACKAGECONFIG[libsolv] = "--with-libsolv,--without-libsolv,libsolv" EXTRA_OECONF:class-native = "--localstatedir=/${@os.path.relpath('${localstatedir}', '${STAGING_DIR_NATIVE}')} --sysconfdir=/${@os.path.relpath('${sysconfdir}', '${STAGING_DIR_NATIVE}')}"--
Alex Stewart
Software Engineer - NI Real-Time OS
NI (National Instruments)-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#170608): https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_message_170608&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=_VWADuvDS3QdSEYWTnh8wNH0lOxVHty18QZynvOtpcU&e=
Mute This Topic: https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_mt_93654146_1997914&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=QncdL2I_pCDsz7AMbsTgStzC7bt5V2x_E_gIblrtjZY&e=
Group Owner: [email protected]
Unsubscribe: https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_unsub&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=PxAE_3yGzbSZ21-LHz_1r3gd2_STPoZcrYDeu9vd_GM&e= [[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
Quoted 8 lines · Alex Feinman
I do have some numbers. When I was selling this change internally, I did a comparison on our internal build.
Combined write IPK times (Σ tdo_package_write_ipk)xz 162m 35s gz 52m 13s zstd 33m 49sCompression rate for zstd was closer to xz than to gz but not as good as xz. For systems that have to cache packages on the device with limited storage xz might be a better option, but for the bulk of projects zstd is the best choice
Additionally, zstd offers much faster decompression than xz so the rootfs build step that includes unpacking all of the ipks, takes 3m 58s with xz and 2m 44s with zstd.
One other thing of note - if your build includes debug packages, some may be quite large. E.g. one of our components produces a 2.2 GB debug package (uncompressed). On large files xz requires a disproportionally large amount of time resulting in 15 minutes needed to simply write ipk for the abovementioned packages, whereas zstd took about 45 sec. For frequent tasks like bitbaking a single package this translates in a lot of saved time.
Those are certainly compelling performance improvements. Assuming that the final data-segment size is within 5%-ish of xz, then I would agree with the rest of the thread that it should probably be the contemporary default.
And if we make it the default compressor for OE IPKs, then obviously my criticism in the original PR is satisfied.
Quoted 1 line
Bottom line - I think making xz a default package compressor was not entirely thought through. gzip or zstd is what the default should be.
ZStandard support was only added to opkg last September [1]. Before that, xz was the new hotness that replaced gzip. :)
[1] https://git.yoctoproject.org/opkg/commit/?id=5dead419e94bce2e6b743ad786c1daec0e1aa294
Quoted 1 line
One final note: I could not find a reasonable explanation for why opkg-tools require code changes to support a different compressor. BSD tar and GNU tar both can easily accept compressors that they have no idea about (via -I option) because all of them provide a unified command line interface for use in pipes. If this were done similar to tar, we could have used any compressor we wanted, including the multithreaded versions (zstdmt)
Well, presumably IPK creation tools can only support the matrix of compression algorithms which your opkg binary can decompress. I suppose someone could try to implement a plugable compression module system for opkg. But given that nearly everyone uses opkg in an embedded context, I'm not sure it would get much use.
Quoted 108 lines · Khem Raj
On Tue, Sep 13, 2022 at 12:19 PM Alex Stewart
<[email protected]> wrote:
>
> ACK from me - apart from enabling zstd by default.
>
> On 9/13/22 07:37, Etienne Cordonnier via lists.openembedded.org
<https://urldefense.com/v3/__http://lists.openembedded.org__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_bxrNjvo$>
wrote:
> > This allows the use of zstd for opkg packages by using
OPKGBUILDCMD:
> > OPKGBUILDCMD = "opkg-build -Z zstd"
> >
> > Signed-off-by: Alex Feinman <[email protected]>
> > Signed-off-by: Etienne Cordonnier <[email protected]>
> > ---
> > meta/recipes-devtools/opkg/opkg_0.6.0.bb
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$>
| 3 ++-
> > 1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/meta/recipes-devtools/opkg/opkg_0.6.0.bb
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$>
b/meta/recipes-devtools/opkg/opkg_0.6.0.bb
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$>
> > index 7b351e8123..e38d9d6f3f 100644
> > --- a/meta/recipes-devtools/opkg/opkg_0.6.0.bb
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$>
> > +++ b/meta/recipes-devtools/opkg/opkg_0.6.0.bb
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$>
> > @@ -30,7 +30,7 @@ inherit autotools pkgconfig ptest
> > target_localstatedir := "${localstatedir}"
> > OPKGLIBDIR ??= "${target_localstatedir}/lib"
> >
> > -PACKAGECONFIG ??= "libsolv"
> > +PACKAGECONFIG ??= "libsolv zstd"
>
> Building in zstd support by default is a little suspect to me.
>
> Unless I'm mistaken, OE-core will only build xz-compressed IPKs by
> default. So zstd support would be unnecessary for a distro
integrator
> who just uses upstream OE-core.
>
> For distros which use zstd compression in their packages, I think it
> would be more appropriate to overwrite the opkg PACKAGECONFIG in a
> .bbappend.
>
This is perhaps fine. I do wonder if there is some performance
comparison data between xz and zstd compressed ipks
with opkg, it might help users on making this choice and also if we
should consider using
zstd by default at some point or not.
> Is there something I'm not considering here?
>
> >
> > PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\
> > gnupg gpgme libgpg-error,\
> > @@ -39,6 +39,7 @@ PACKAGECONFIG[gpg] =
"--enable-gpg,--disable-gpg,\
> > PACKAGECONFIG[curl] = "--enable-curl,--disable-curl,curl"
> > PACKAGECONFIG[ssl-curl] =
"--enable-ssl-curl,--disable-ssl-curl,curl openssl"
> > PACKAGECONFIG[sha256] = "--enable-sha256,--disable-sha256"
> > +PACKAGECONFIG[zstd] = "--enable-zstd,--disable-zstd,zstd"
> > PACKAGECONFIG[libsolv] =
"--with-libsolv,--without-libsolv,libsolv"
> >
> > EXTRA_OECONF:class-native =
"--localstatedir=/${@os.path.relpath('${localstatedir}',
'${STAGING_DIR_NATIVE}')}
--sysconfdir=/${@os.path.relpath('${sysconfdir}',
'${STAGING_DIR_NATIVE}')}"
> >
> >
> >
>
> --
> Alex Stewart
> Software Engineer - NI Real-Time OS
> NI (National Instruments)
>
> [email protected]
>
>
> -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> View/Reply Online (#170608):
https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_message_170608&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=_VWADuvDS3QdSEYWTnh8wNH0lOxVHty18QZynvOtpcU&e=
<https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_message_170608&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=_VWADuvDS3QdSEYWTnh8wNH0lOxVHty18QZynvOtpcU&e=>
> Mute This Topic:
https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_mt_93654146_1997914&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=QncdL2I_pCDsz7AMbsTgStzC7bt5V2x_E_gIblrtjZY&e=
<https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_mt_93654146_1997914&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=QncdL2I_pCDsz7AMbsTgStzC7bt5V2x_E_gIblrtjZY&e=>
> Group Owner: [email protected]
<mailto:openembedded-core%[email protected]>
> Unsubscribe:
https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_unsub&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=PxAE_3yGzbSZ21-LHz_1r3gd2_STPoZcrYDeu9vd_GM&e=
<https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_unsub&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=PxAE_3yGzbSZ21-LHz_1r3gd2_STPoZcrYDeu9vd_GM&e=>
[[email protected]]
> -=-=-=-=-=-=-=-=-=-=-=-
>Signature · 5 lines
Alex Stewart Software Engineer - NI Real-Time OS NI (National Instruments) [email protected]
I ran a build of core-image-full-cmdline using xz and zstd, using pre-populated downloads and sstate-cache directories but with empty tmp directory. Here are the numbers:
zstd:
bitbake core-image-full-cmdline took 2m52.768s (real), the resulting directory tmp/deploy/ipk is 1.6GB big.
xz:
bitbake core-image-full-cmdline took 4m14.214s (real), the resulting directory tmp/deploy/ipk/ is 996M big
So the build with xz is 47% slower (254/172) than with zstd for this "incremental build" use-case.
See the 5 biggest packages, the difference in compression-ratio increases with big files:
build$ ls -lh tmp-zstd/deploy/ipk/core2-64/ --sort=size | head -n5 total 1.6G
-rw-r--r-- 3 ecordonnier ecordonnier 331M Sep 14 10:39 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 264M Sep 14 10:39 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 113M Sep 14 10:39 openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 47M Sep 14 10:39
binutils-dbg_2.39-r0_core2-64.ipk
build$ ls -lh tmp-xz/deploy/ipk/core2-64/ --sort=size | head -n5 total 963M
-rw-r--r-- 3 ecordonnier ecordonnier 214M Sep 14 10:44 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 193M Sep 14 10:44 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 35M Sep 14 10:44
openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 32M Sep 14 10:44
binutils-dbg_2.39-r0_core2-64.ipk
...
I think for use-cases where the size of the ipk packages matters, xz is better. For use-cases where it does not matter (ipk packages not deployed), build-time matters more than compression-ratio and zstd is better.
Regarding the enablement of zstd in opkg per default, I'll send a new version of the patch without this line.
My thinking for enabling zstd per default in opkg was that zstd is already enabled per default in libarchive's PACKAGECONFIG, and disabling zstd in opkg's PACKAGECONFIG removes only a few lines of code from opkg (opkg uses libarchive for doing the actual compression/decompression).
Quoted 196 lines · Alex Stewart
Quoted 6 lines · Alex Feinman
I do have some numbers. When I was selling this change internally, I did a comparison on our internal build.
Combined write IPK times (Σ tdo_package_write_ipk)xz 162m 35s gz 52m 13s zstd 33m 49sQuoted 14 lines
Compression rate for zstd was closer to xz than to gz but not as good as xz. For systems that have to cache packages on the device with limited storage xz might be a better option, but for the bulk of projects zstd is the best choice
Additionally, zstd offers much faster decompression than xz so the rootfs build step that includes unpacking all of the ipks, takes 3m 58s with xz and 2m 44s with zstd.
One other thing of note - if your build includes debug packages, some may be quite large. E.g. one of our components produces a 2.2 GB debug package (uncompressed). On large files xz requires a disproportionally large amount of time resulting in 15 minutes needed to simply write ipk for the abovementioned packages, whereas zstd took about 45 sec. For frequent tasks like bitbaking a single package this translates in a lot of saved time.Those are certainly compelling performance improvements. Assuming that the final data-segment size is within 5%-ish of xz, then I would agree with the rest of the thread that it should probably be the contemporary default.
And if we make it the default compressor for OE IPKs, then obviously my criticism in the original PR is satisfied.
Quoted 2 lines
Bottom line - I think making xz a default package compressor was not entirely thought through. gzip or zstd is what the default should be.
ZStandard support was only added to opkg last September [1]. Before that, xz was the new hotness that replaced gzip. :)
[1]
Quoted 7 lines
One final note: I could not find a reasonable explanation for why opkg-tools require code changes to support a different compressor. BSD tar and GNU tar both can easily accept compressors that they have no idea about (via -I option) because all of them provide a unified command line interface for use in pipes. If this were done similar to tar, we could have used any compressor we wanted, including the multithreaded versions (zstdmt)
Well, presumably IPK creation tools can only support the matrix of compression algorithms which your opkg binary can decompress. I suppose someone could try to implement a plugable compression module system for opkg. But given that nearly everyone uses opkg in an embedded context, I'm not sure it would get much use.
Quoted 10 lines · Khem Raj
On Tue, Sep 13, 2022 at 12:19 PM Alex Stewart <[email protected]> wrote: > > ACK from me - apart from enabling zstd by default. > > On 9/13/22 07:37, Etienne Cordonnier via lists.openembedded.org <Quoted 11 lines
wrote:
> > This allows the use of zstd for opkg packages by using OPKGBUILDCMD:
> > OPKGBUILDCMD = "opkg-build -Z zstd"
> >
> > Signed-off-by: Alex Feinman <[email protected]> > > Signed-off-by: Etienne Cordonnier <[email protected]> > > ---
> > meta/recipes-devtools/opkg/opkg_0.6.0.bb
<Quoted 6 lines
| 3 ++- > > 1 file changed, 2 insertions(+), 1 deletion(-) > > > > diff --git a/meta/recipes-devtools/opkg/opkg_0.6.0.bb <Quoted 3 lines
b/meta/recipes-devtools/opkg/opkg_0.6.0.bb <Quoted 4 lines
Quoted 3 lines
Quoted 3 lines
index
7b351e8123..e38d9d6f3f100644
--- a/meta/recipes-devtools/opkg/opkg_0.6.0.bb<Quoted 3 lines
Quoted 2 lines
Quoted 2 lines
+++ b/meta/recipes-devtools/opkg/opkg_0.6.0.bb
<Quoted 10 lines
Quoted 10 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb inherit autotools pkgconfig ptest · line 33
Quoted 7 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb +1
-1@@ -30,7 +30,7 @@ inherit autotools pkgconfig ptest target_localstatedir := "${localstatedir}" OPKGLIBDIR ??= "${target_localstatedir}/lib" -PACKAGECONFIG ??= "libsolv" +PACKAGECONFIG ??= "libsolv zstd"Building in zstd support by default is a little suspect to me.
Quoted 6 lines
Quoted 2 lines
Unless I'm mistaken, OE-core will only build xz-compressed IPKs by default. So zstd support would be unnecessary for a distro
integrator
> who just uses upstream OE-core.
>
> For distros which use zstd compression in their packages, I thinkit
Quoted 37 lines
Quoted 3 lines
would be more appropriate to overwrite the opkg PACKAGECONFIG in a .bbappend.
This is perhaps fine. I do wonder if there is some performance comparison data between xz and zstd compressed ipks with opkg, it might help users on making this choice and also if we should consider using zstd by default at some point or not. > Is there something I'm not considering here? > > > > > PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ > > gnupg gpgme libgpg-error,\ > > @@ -39,6 +39,7 @@ PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ > > PACKAGECONFIG[curl] = "--enable-curl,--disable-curl,curl" > > PACKAGECONFIG[ssl-curl] = "--enable-ssl-curl,--disable-ssl-curl,curl openssl" > > PACKAGECONFIG[sha256] = "--enable-sha256,--disable-sha256" > > +PACKAGECONFIG[zstd] = "--enable-zstd,--disable-zstd,zstd" > > PACKAGECONFIG[libsolv] = "--with-libsolv,--without-libsolv,libsolv" > > > > EXTRA_OECONF:class-native = "--localstatedir=/${@os.path.relpath('${localstatedir}', '${STAGING_DIR_NATIVE}')} --sysconfdir=/${@os.path.relpath('${sysconfdir}', '${STAGING_DIR_NATIVE}')}" > > > > > > > > -- > Alex Stewart > Software Engineer - NI Real-Time OSQuoted 9 lines
> NI (National Instruments) > > [email protected] > > > -=-=-=-=-=-=-=-=-=-=-=- > Links: You receive all messages sent to this group. > View/Reply Online (#170608):Quoted 1 line
<Quoted 4 lines
Quoted 3 lines
Mute This Topic:
Quoted 1 line
<Quoted 6 lines
Quoted 3 lines
Group Owner: [email protected]
<mailto:openembedded-core%[email protected]> > Unsubscribe:
Quoted 1 line
<Quoted 5 lines
[[email protected]] > -=-=-=-=-=-=-=-=-=-=-=- >--
Alex Stewart
Software Engineer - NI Real-Time OS
NI (National Instruments)
Also note that zstd's default compression level is 3 per default (from a 1 to 19 scale). I did not test other compression levels.
Quoted 246 lines · Etienne Cordonnier
I ran a build of core-image-full-cmdline using xz and zstd, using pre-populated downloads and sstate-cache directories but with empty tmp directory. Here are the numbers:
zstd:
bitbake core-image-full-cmdline took 2m52.768s (real), the resulting directory tmp/deploy/ipk is 1.6GB big.
xz:
bitbake core-image-full-cmdline took 4m14.214s (real), the resulting directory tmp/deploy/ipk/ is 996M bigSo the build with xz is 47% slower (254/172) than with zstd for this "incremental build" use-case.
See the 5 biggest packages, the difference in compression-ratio increases with big files:
build$ ls -lh tmp-zstd/deploy/ipk/core2-64/ --sort=size | head -n5 total 1.6G
-rw-r--r-- 3 ecordonnier ecordonnier 331M Sep 14 10:39 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 264M Sep 14 10:39 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 113M Sep 14 10:39 openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 47M Sep 14 10:39
binutils-dbg_2.39-r0_core2-64.ipk
build$ ls -lh tmp-xz/deploy/ipk/core2-64/ --sort=size | head -n5 total 963M
-rw-r--r-- 3 ecordonnier ecordonnier 214M Sep 14 10:44 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 193M Sep 14 10:44 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 35M Sep 14 10:44
openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 32M Sep 14 10:44
binutils-dbg_2.39-r0_core2-64.ipk
...I think for use-cases where the size of the ipk packages matters, xz is better. For use-cases where it does not matter (ipk packages not deployed), build-time matters more than compression-ratio and zstd is better.
Regarding the enablement of zstd in opkg per default, I'll send a new version of the patch without this line.
My thinking for enabling zstd per default in opkg was that zstd is already enabled per default in libarchive's PACKAGECONFIG, and disabling zstd in opkg's PACKAGECONFIG removes only a few lines of code from opkg (opkg uses libarchive for doing the actual compression/decompression).Quoted 196 lines · Alex Stewart
Quoted 6 lines · Alex Feinman
I do have some numbers. When I was selling this change internally, I did a comparison on our internal build.
Combined write IPK times (Σ tdo_package_write_ipk)xz 162m 35s gz 52m 13s zstd 33m 49sQuoted 14 lines
Compression rate for zstd was closer to xz than to gz but not as good as xz. For systems that have to cache packages on the device with limited storage xz might be a better option, but for the bulk of projects zstd is the best choice
Additionally, zstd offers much faster decompression than xz so the rootfs build step that includes unpacking all of the ipks, takes 3m 58s with xz and 2m 44s with zstd.
One other thing of note - if your build includes debug packages, some may be quite large. E.g. one of our components produces a 2.2 GB debug package (uncompressed). On large files xz requires a disproportionally large amount of time resulting in 15 minutes needed to simply write ipk for the abovementioned packages, whereas zstd took about 45 sec. For frequent tasks like bitbaking a single package this translates in a lot of saved time.Those are certainly compelling performance improvements. Assuming that the final data-segment size is within 5%-ish of xz, then I would agree with the rest of the thread that it should probably be the contemporary default.
And if we make it the default compressor for OE IPKs, then obviously my criticism in the original PR is satisfied.
Quoted 2 lines
Bottom line - I think making xz a default package compressor was not entirely thought through. gzip or zstd is what the default should be.
ZStandard support was only added to opkg last September [1]. Before that, xz was the new hotness that replaced gzip. :)
[1]
Quoted 7 lines
One final note: I could not find a reasonable explanation for why opkg-tools require code changes to support a different compressor. BSD tar and GNU tar both can easily accept compressors that they have no idea about (via -I option) because all of them provide a unified command line interface for use in pipes. If this were done similar to tar, we could have used any compressor we wanted, including the multithreaded versions (zstdmt)
Well, presumably IPK creation tools can only support the matrix of compression algorithms which your opkg binary can decompress. I suppose someone could try to implement a plugable compression module system for opkg. But given that nearly everyone uses opkg in an embedded context, I'm not sure it would get much use.
Quoted 10 lines · Khem Raj
On Tue, Sep 13, 2022 at 12:19 PM Alex Stewart <[email protected]> wrote: > > ACK from me - apart from enabling zstd by default. > > On 9/13/22 07:37, Etienne Cordonnier via lists.openembedded.org <Quoted 11 lines
wrote:
> > This allows the use of zstd for opkg packages by using OPKGBUILDCMD:
> > OPKGBUILDCMD = "opkg-build -Z zstd"
> >
> > Signed-off-by: Alex Feinman <[email protected]> > > Signed-off-by: Etienne Cordonnier <[email protected]> > > ---
> > meta/recipes-devtools/opkg/opkg_0.6.0.bb
<Quoted 6 lines
| 3 ++- > > 1 file changed, 2 insertions(+), 1 deletion(-) > > > > diff --git a/meta/recipes-devtools/opkg/opkg_0.6.0.bb <Quoted 3 lines
b/meta/recipes-devtools/opkg/opkg_0.6.0.bb <Quoted 4 lines
Quoted 3 lines
Quoted 3 lines
index
7b351e8123..e38d9d6f3f100644
--- a/meta/recipes-devtools/opkg/opkg_0.6.0.bb<Quoted 3 lines
Quoted 2 lines
Quoted 2 lines
+++ b/meta/recipes-devtools/opkg/opkg_0.6.0.bb
<Quoted 10 lines
Quoted 10 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb inherit autotools pkgconfig ptest · line 33
Quoted 7 lines
meta/recipes-devtools/opkg/opkg_0.6.0.bb +1
-1@@ -30,7 +30,7 @@ inherit autotools pkgconfig ptest target_localstatedir := "${localstatedir}" OPKGLIBDIR ??= "${target_localstatedir}/lib" -PACKAGECONFIG ??= "libsolv" +PACKAGECONFIG ??= "libsolv zstd"Building in zstd support by default is a little suspect to me.
Quoted 6 lines
Quoted 2 lines
Unless I'm mistaken, OE-core will only build xz-compressed IPKs by default. So zstd support would be unnecessary for a distro
integrator
> who just uses upstream OE-core.
>
> For distros which use zstd compression in their packages, I thinkit
Quoted 37 lines
Quoted 3 lines
would be more appropriate to overwrite the opkg PACKAGECONFIG in a .bbappend.
This is perhaps fine. I do wonder if there is some performance comparison data between xz and zstd compressed ipks with opkg, it might help users on making this choice and also if we should consider using zstd by default at some point or not. > Is there something I'm not considering here? > > > > > PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ > > gnupg gpgme libgpg-error,\ > > @@ -39,6 +39,7 @@ PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ > > PACKAGECONFIG[curl] = "--enable-curl,--disable-curl,curl" > > PACKAGECONFIG[ssl-curl] = "--enable-ssl-curl,--disable-ssl-curl,curl openssl" > > PACKAGECONFIG[sha256] = "--enable-sha256,--disable-sha256" > > +PACKAGECONFIG[zstd] = "--enable-zstd,--disable-zstd,zstd" > > PACKAGECONFIG[libsolv] = "--with-libsolv,--without-libsolv,libsolv" > > > > EXTRA_OECONF:class-native = "--localstatedir=/${@os.path.relpath('${localstatedir}', '${STAGING_DIR_NATIVE}')} --sysconfdir=/${@os.path.relpath('${sysconfdir}', '${STAGING_DIR_NATIVE}')}" > > > > > > > > -- > Alex Stewart > Software Engineer - NI Real-Time OSQuoted 9 lines
> NI (National Instruments) > > [email protected] > > > -=-=-=-=-=-=-=-=-=-=-=- > Links: You receive all messages sent to this group. > View/Reply Online (#170608):Quoted 1 line
<Quoted 4 lines
Quoted 3 lines
Mute This Topic:
Quoted 1 line
<Quoted 6 lines
Quoted 3 lines
Group Owner: [email protected]
<mailto:openembedded-core%[email protected]> > Unsubscribe:
Quoted 1 line
<Quoted 5 lines
[[email protected]] > -=-=-=-=-=-=-=-=-=-=-=- >--
Alex Stewart
Software Engineer - NI Real-Time OS
NI (National Instruments)
Thanks for checking.
I'd be interested to know if setting a higher compression level for zstd can get us to a similar compression ratio to xz. If so, then I think it could be some real value to distro maintainers to be able to *tune* their compression.
That's not blocking for your new PR though.
Quoted 297 lines · Etienne Cordonnier
Also note that zstd's default compression level is 3 per default (from a 1 to 19 scale). I did not test other compression levels.
On Wed, Sep 14, 2022 at 11:58 AM Etienne Cordonnier <[email protected]> wrote:
I ran a build of core-image-full-cmdline using xz and zstd, using pre-populated downloads and sstate-cache directories but with empty tmp directory. Here are the numbers:
zstd:
bitbake core-image-full-cmdline took 2m52.768s (real), the resulting directory tmp/deploy/ipk is 1.6GB big.
xz:
bitbake core-image-full-cmdline took 4m14.214s (real), the resulting directory tmp/deploy/ipk/ is 996M bigSo the build with xz is 47% slower (254/172) than with zstd for this "incremental build" use-case.
See the 5 biggest packages, the difference in compression-ratio increases with big files:
build$ ls -lh tmp-zstd/deploy/ipk/core2-64/ --sort=size | head -n5 total 1.6G
-rw-r--r-- 3 ecordonnier ecordonnier 331M Sep 14 10:39 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 264M Sep 14 10:39 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 113M Sep 14 10:39 openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 47M Sep 14 10:39 binutils-dbg_2.39-r0_core2-64.ipk
build$ ls -lh tmp-xz/deploy/ipk/core2-64/ --sort=size | head -n5 total 963M
-rw-r--r-- 3 ecordonnier ecordonnier 214M Sep 14 10:44 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 193M Sep 14 10:44 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 35M Sep 14 10:44 openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 32M Sep 14 10:44 binutils-dbg_2.39-r0_core2-64.ipk
...I think for use-cases where the size of the ipk packages matters, xz is better. For use-cases where it does not matter (ipk packages not deployed), build-time matters more than compression-ratio and zstd is better.
Regarding the enablement of zstd in opkg per default, I'll send a new version of the patch without this line.
My thinking for enabling zstd per default in opkg was that zstd is already enabled per default in libarchive's PACKAGECONFIG, and disabling zstd in opkg's PACKAGECONFIG removes only a few lines of code from opkg (opkg uses libarchive for doing the actual compression/decompression).On Tue, Sep 13, 2022 at 11:57 PM Alex Stewart
<[email protected]> wrote:On 9/13/22 15:20, Alex Feinman wrote:
> I do have some numbers. When I was selling this change internally, I
> did a comparison on our internal build.
> Combined write IPK times (Σ tdo_package_write_ipk) > xz 162m 35s
> gz 52m 13s
> zstd 33m 49s
> Compression rate for zstd was closer to xz than to gz but not as good
> as xz. For systems that have to cache packages on the device with
> limited storage xz might be a better option, but for the bulk of
> projects zstd is the best choice
> Additionally, zstd offers much faster decompression than xz so the
> rootfs build step that includes unpacking all of the ipks, takes 3m
> 58s with xz and 2m 44s with zstd.
> One other thing of note - if your build includes debug packages, some
> may be quite large. E.g. one of our components produces a 2.2 GB debug
> package (uncompressed). On large files xz requires a disproportionally
> large amount of time resulting in 15 minutes needed to simply write
> ipk for the abovementioned packages, whereas zstd took about 45 sec.
> For frequent tasks like bitbaking a single package this translates in
> a lot of saved time.Those are certainly compelling performance improvements. Assuming that
the final data-segment size is within 5%-ish of xz, then I would agree
with the rest of the thread that it should probably be the contemporary
default.And if we make it the default compressor for OE IPKs, then obviously my
criticism in the original PR is satisfied.> Bottom line - I think making xz a default package compressor was not
> entirely thought through. gzip or zstd is what the default should be.ZStandard support was only added to opkg last September [1]. Before
that, xz was the new hotness that replaced gzip. :)[1]
https://urldefense.proofpoint.com/v2/url?u=https-3A__git.yoctoproject.org_opkg_commit_-3Fid-3D5dead419e94bce2e6b743ad786c1daec0e1aa294&d=DwIDaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=AhkbNonVuMIGRfPx_Qj9TsyDLWdbBqarUzFxz3aALck&m=VG_fgpCGq8Zgu73KQP1wTtb1D1TZftNpXj8jGnouPtGIBYyLIZbG8F-85POUcVN7&s=Mwoq2kkjfZhto6J5OomduJ5Rhyg_oSe-dkjeltE4Ls8&e=
<https://urldefense.proofpoint.com/v2/url?u=https-3A__git.yoctoproject.org_opkg_commit_-3Fid-3D5dead419e94bce2e6b743ad786c1daec0e1aa294&d=DwIDaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=AhkbNonVuMIGRfPx_Qj9TsyDLWdbBqarUzFxz3aALck&m=VG_fgpCGq8Zgu73KQP1wTtb1D1TZftNpXj8jGnouPtGIBYyLIZbG8F-85POUcVN7&s=Mwoq2kkjfZhto6J5OomduJ5Rhyg_oSe-dkjeltE4Ls8&e=>> One final note: I could not find a reasonable explanation for why
> opkg-tools require code changes to support a different compressor. BSD
> tar and GNU tar both can easily accept compressors that they have no
> idea about (via -I option) because all of them provide a unified
> command line interface for use in pipes. If this were done similar to
> tar, we could have used any compressor we wanted, including the > multithreaded versions (zstdmt)Well, presumably IPK creation tools can only support the matrix of
compression algorithms which your opkg binary can decompress. I suppose
someone could try to implement a plugable compression module system for
opkg. But given that nearly everyone uses opkg in an embedded context,
I'm not sure it would get much use.>
> On Tue, Sep 13, 2022 at 12:43 PM Khem Raj <[email protected]> wrote:
>
> On Tue, Sep 13, 2022 at 12:19 PM Alex Stewart
> <[email protected]> wrote:
> >
> > ACK from me - apart from enabling zstd by default. > >
> > On 9/13/22 07:37, Etienne Cordonnier via lists.openembedded.org
<https://urldefense.com/v3/__http://lists.openembedded.org__;!!FbZ0ZwI3Qg!t0JG8eQqpsPs8clofb2phwJc6ONCKBjy-iQukKenLRVzBQvJrKTp8VsgKBI7cR8KlV3N64C3fKRHIcFhOF1_vg$>
>
<https://urldefense.com/v3/__http://lists.openembedded.org__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_bxrNjvo$>
> wrote:
> > > This allows the use of zstd for opkg packages by using > OPKGBUILDCMD:
> > > OPKGBUILDCMD = "opkg-build -Z zstd"
> > >
> > > Signed-off-by: Alex Feinman <[email protected]> > > > Signed-off-by: Etienne Cordonnier <[email protected]> > > > ---
> > > meta/recipes-devtools/opkg/opkg_0.6.0.bb
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!t0JG8eQqpsPs8clofb2phwJc6ONCKBjy-iQukKenLRVzBQvJrKTp8VsgKBI7cR8KlV3N64C3fKRHIcFFmbls2A$>
>
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$>
> | 3 ++-
> > > 1 file changed, 2 insertions(+), 1 deletion(-) > > >
> > > diff --git a/meta/recipes-devtools/opkg/opkg_0.6.0.bb
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!t0JG8eQqpsPs8clofb2phwJc6ONCKBjy-iQukKenLRVzBQvJrKTp8VsgKBI7cR8KlV3N64C3fKRHIcFFmbls2A$>
>
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$>
> b/meta/recipes-devtools/opkg/opkg_0.6.0.bb
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!t0JG8eQqpsPs8clofb2phwJc6ONCKBjy-iQukKenLRVzBQvJrKTp8VsgKBI7cR8KlV3N64C3fKRHIcFFmbls2A$>
>
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$>
> > > index7b351e8123..e38d9d6f3f100644
> > > --- a/meta/recipes-devtools/opkg/opkg_0.6.0.bb
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!t0JG8eQqpsPs8clofb2phwJc6ONCKBjy-iQukKenLRVzBQvJrKTp8VsgKBI7cR8KlV3N64C3fKRHIcFFmbls2A$>
>
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$>
> > > +++ b/meta/recipes-devtools/opkg/opkg_0.6.0.bb
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!t0JG8eQqpsPs8clofb2phwJc6ONCKBjy-iQukKenLRVzBQvJrKTp8VsgKBI7cR8KlV3N64C3fKRHIcFFmbls2A$>
>
<https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$>
> > > @@ -30,7 +30,7 @@ inherit autotools pkgconfig ptest > > > target_localstatedir := "${localstatedir}" > > > OPKGLIBDIR ??= "${target_localstatedir}/lib" > > >
> > > -PACKAGECONFIG ??= "libsolv"
> > > +PACKAGECONFIG ??= "libsolv zstd"
> >
> > Building in zstd support by default is a little suspect to me.
> >
> > Unless I'm mistaken, OE-core will only build xz-compressed IPKs by
> > default. So zstd support would be unnecessary for a distro > integrator
> > who just uses upstream OE-core.
> >
> > For distros which use zstd compression in their packages, I think it
> > would be more appropriate to overwrite the opkg PACKAGECONFIG in a
> > .bbappend.
> >
>
> This is perhaps fine. I do wonder if there is some performance
> comparison data between xz and zstd compressed ipks > with opkg, it might help users on making this choice and also if we
> should consider using
> zstd by default at some point or not.
>
> > Is there something I'm not considering here?
> >
> > >
> > > PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\
> > > gnupg gpgme libgpg-error,\
> > > @@ -39,6 +39,7 @@ PACKAGECONFIG[gpg] =
> "--enable-gpg,--disable-gpg,\
> > > PACKAGECONFIG[curl] = "--enable-curl,--disable-curl,curl"
> > > PACKAGECONFIG[ssl-curl] =
> "--enable-ssl-curl,--disable-ssl-curl,curl openssl" > > > PACKAGECONFIG[sha256] = "--enable-sha256,--disable-sha256"
> > > +PACKAGECONFIG[zstd] = "--enable-zstd,--disable-zstd,zstd"
> > > PACKAGECONFIG[libsolv] =
> "--with-libsolv,--without-libsolv,libsolv"
> > >
> > > EXTRA_OECONF:class-native =
> "--localstatedir=/${@os.path.relpath('${localstatedir}', > '${STAGING_DIR_NATIVE}')}
> --sysconfdir=/${@os.path.relpath('${sysconfdir}', > '${STAGING_DIR_NATIVE}')}"
> > >
> > >
> > >
> >
> > --
> > Alex Stewart
> > Software Engineer - NI Real-Time OS
> > NI (National Instruments)
> >
> > [email protected]
> >
> >
> > -=-=-=-=-=-=-=-=-=-=-=-
> > Links: You receive all messages sent to this group. > > View/Reply Online (#170608):
>
https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_message_170608&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=_VWADuvDS3QdSEYWTnh8wNH0lOxVHty18QZynvOtpcU&e=
<https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_message_170608&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=_VWADuvDS3QdSEYWTnh8wNH0lOxVHty18QZynvOtpcU&e=>
>
<https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_message_170608&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=_VWADuvDS3QdSEYWTnh8wNH0lOxVHty18QZynvOtpcU&e=
<https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_message_170608&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=_VWADuvDS3QdSEYWTnh8wNH0lOxVHty18QZynvOtpcU&e=>>
>
> > Mute This Topic:
>
https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_mt_93654146_1997914&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=QncdL2I_pCDsz7AMbsTgStzC7bt5V2x_E_gIblrtjZY&e=
<https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_mt_93654146_1997914&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=QncdL2I_pCDsz7AMbsTgStzC7bt5V2x_E_gIblrtjZY&e=>
>
<https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_mt_93654146_1997914&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=QncdL2I_pCDsz7AMbsTgStzC7bt5V2x_E_gIblrtjZY&e=
<https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_mt_93654146_1997914&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=QncdL2I_pCDsz7AMbsTgStzC7bt5V2x_E_gIblrtjZY&e=>>
>
> > Group Owner:
[email protected] <mailto:openembedded-core%[email protected]> > <mailto:openembedded-core%[email protected] <mailto:openembedded-core%[email protected]>> > > Unsubscribe:
>
https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_unsub&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=PxAE_3yGzbSZ21-LHz_1r3gd2_STPoZcrYDeu9vd_GM&e=
<https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_unsub&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=PxAE_3yGzbSZ21-LHz_1r3gd2_STPoZcrYDeu9vd_GM&e=>
>
<https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_unsub&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=PxAE_3yGzbSZ21-LHz_1r3gd2_STPoZcrYDeu9vd_GM&e=
<https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_unsub&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=PxAE_3yGzbSZ21-LHz_1r3gd2_STPoZcrYDeu9vd_GM&e=>>
> [[email protected]]
> > -=-=-=-=-=-=-=-=-=-=-=-
> >
>--
Alex Stewart
Software Engineer - NI Real-Time OS
NI (National Instruments)
Signature · 5 lines
Alex Stewart Software Engineer - NI Real-Time OS NI (National Instruments) [email protected]
Quoted 7 lines · Alex Stewart
Thanks for checking.
I'd be interested to know if setting a higher compression level for zstd can get us to a similar compression ratio to xz. If so, then I think it could be some real value to distro maintainers to be able to *tune* their compression.
yeah it will be interesting to say try something like level 9 but I think times might regress with that but it might be good to know the balance and perhaps suggest size mode and performance mode of zstd instead of xz
Quoted 312 lines
That's not blocking for your new PR though.
Quoted 290 lines · Etienne Cordonnier
Also note that zstd's default compression level is 3 per default (from a 1 to 19 scale). I did not test other compression levels.
On Wed, Sep 14, 2022 at 11:58 AM Etienne Cordonnier <[email protected]> wrote:
I ran a build of core-image-full-cmdline using xz and zstd, using pre-populated downloads and sstate-cache directories but with empty tmp directory. Here are the numbers: zstd: bitbake core-image-full-cmdline took 2m52.768s (real), the resulting directory tmp/deploy/ipk is 1.6GB big. xz: bitbake core-image-full-cmdline took 4m14.214s (real), the resulting directory tmp/deploy/ipk/ is 996M big So the build with xz is 47% slower (254/172) than with zstd for this "incremental build" use-case. See the 5 biggest packages, the difference in compression-ratio increases with big files: build$ ls -lh tmp-zstd/deploy/ipk/core2-64/ --sort=size | head -n5 total 1.6G -rw-r--r-- 3 ecordonnier ecordonnier 331M Sep 14 10:39 gcc-dbg_12.2.0-r0_core2-64.ipk -rw-r--r-- 3 ecordonnier ecordonnier 264M Sep 14 10:39 openssl-dbg_3.0.5-r0_core2-64.ipk -rw-r--r-- 3 ecordonnier ecordonnier 113M Sep 14 10:39 openssl-ptest_3.0.5-r0_core2-64.ipk -rw-r--r-- 3 ecordonnier ecordonnier 47M Sep 14 10:39 binutils-dbg_2.39-r0_core2-64.ipk build$ ls -lh tmp-xz/deploy/ipk/core2-64/ --sort=size | head -n5 total 963M -rw-r--r-- 3 ecordonnier ecordonnier 214M Sep 14 10:44 gcc-dbg_12.2.0-r0_core2-64.ipk -rw-r--r-- 3 ecordonnier ecordonnier 193M Sep 14 10:44 openssl-dbg_3.0.5-r0_core2-64.ipk -rw-r--r-- 3 ecordonnier ecordonnier 35M Sep 14 10:44 openssl-ptest_3.0.5-r0_core2-64.ipk -rw-r--r-- 3 ecordonnier ecordonnier 32M Sep 14 10:44 binutils-dbg_2.39-r0_core2-64.ipk ... I think for use-cases where the size of the ipk packages matters, xz is better. For use-cases where it does not matter (ipk packages not deployed), build-time matters more than compression-ratio and zstd is better. Regarding the enablement of zstd in opkg per default, I'll send a new version of the patch without this line. My thinking for enabling zstd per default in opkg was that zstd is already enabled per default in libarchive's PACKAGECONFIG, and disabling zstd in opkg's PACKAGECONFIG removes only a few lines of code from opkg (opkg uses libarchive for doing the actual compression/decompression). On Tue, Sep 13, 2022 at 11:57 PM Alex Stewart <[email protected]> wrote: On 9/13/22 15:20, Alex Feinman wrote: > I do have some numbers. When I was selling this change internally, I > did a comparison on our internal build. > Combined write IPK times (Σ t do_package_write_ipk) > xz 162m 35s > gz 52m 13s > zstd 33m 49s > Compression rate for zstd was closer to xz than to gz but not as good > as xz. For systems that have to cache packages on the device with > limited storage xz might be a better option, but for the bulk of > projects zstd is the best choice > Additionally, zstd offers much faster decompression than xz so the > rootfs build step that includes unpacking all of the ipks, takes 3m > 58s with xz and 2m 44s with zstd. > One other thing of note - if your build includes debug packages, some > may be quite large. E.g. one of our components produces a 2.2 GB debug > package (uncompressed). On large files xz requires a disproportionally > large amount of time resulting in 15 minutes needed to simply write > ipk for the abovementioned packages, whereas zstd took about 45 sec. > For frequent tasks like bitbaking a single package this translates in > a lot of saved time. Those are certainly compelling performance improvements. Assuming that the final data-segment size is within 5%-ish of xz, then I would agree with the rest of the thread that it should probably be the contemporary default. And if we make it the default compressor for OE IPKs, then obviously my criticism in the original PR is satisfied. > Bottom line - I think making xz a default package compressor was not > entirely thought through. gzip or zstd is what the default should be. ZStandard support was only added to opkg last September [1]. Before that, xz was the new hotness that replaced gzip. :) [1] https://urldefense.proofpoint.com/v2/url?u=https-3A__git.yoctoproject.org_opkg_commit_-3Fid-3D5dead419e94bce2e6b743ad786c1daec0e1aa294&d=DwIDaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=AhkbNonVuMIGRfPx_Qj9TsyDLWdbBqarUzFxz3aALck&m=VG_fgpCGq8Zgu73KQP1wTtb1D1TZftNpXj8jGnouPtGIBYyLIZbG8F-85POUcVN7&s=Mwoq2kkjfZhto6J5OomduJ5Rhyg_oSe-dkjeltE4Ls8&e= <https://urldefense.proofpoint.com/v2/url?u=https-3A__git.yoctoproject.org_opkg_commit_-3Fid-3D5dead419e94bce2e6b743ad786c1daec0e1aa294&d=DwIDaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=AhkbNonVuMIGRfPx_Qj9TsyDLWdbBqarUzFxz3aALck&m=VG_fgpCGq8Zgu73KQP1wTtb1D1TZftNpXj8jGnouPtGIBYyLIZbG8F-85POUcVN7&s=Mwoq2kkjfZhto6J5OomduJ5Rhyg_oSe-dkjeltE4Ls8&e=> > One final note: I could not find a reasonable explanation for why > opkg-tools require code changes to support a different compressor. BSD > tar and GNU tar both can easily accept compressors that they have no > idea about (via -I option) because all of them provide a unified > command line interface for use in pipes. If this were done similar to > tar, we could have used any compressor we wanted, including the > multithreaded versions (zstdmt) Well, presumably IPK creation tools can only support the matrix of compression algorithms which your opkg binary can decompress. I suppose someone could try to implement a plugable compression module system for opkg. But given that nearly everyone uses opkg in an embedded context, I'm not sure it would get much use. > > On Tue, Sep 13, 2022 at 12:43 PM Khem Raj <[email protected]> wrote: > > On Tue, Sep 13, 2022 at 12:19 PM Alex Stewart > <[email protected]> wrote: > > > > ACK from me - apart from enabling zstd by default. > > > > On 9/13/22 07:37, Etienne Cordonnier via lists.openembedded.org <https://urldefense.com/v3/__http://lists.openembedded.org__;!!FbZ0ZwI3Qg!t0JG8eQqpsPs8clofb2phwJc6ONCKBjy-iQukKenLRVzBQvJrKTp8VsgKBI7cR8KlV3N64C3fKRHIcFhOF1_vg$> > <https://urldefense.com/v3/__http://lists.openembedded.org__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_bxrNjvo$> > wrote: > > > This allows the use of zstd for opkg packages by using > OPKGBUILDCMD: > > > OPKGBUILDCMD = "opkg-build -Z zstd" > > > > > > Signed-off-by: Alex Feinman <[email protected]> > > > Signed-off-by: Etienne Cordonnier <[email protected]> > > > --- > > > meta/recipes-devtools/opkg/opkg_0.6.0.bb <https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!t0JG8eQqpsPs8clofb2phwJc6ONCKBjy-iQukKenLRVzBQvJrKTp8VsgKBI7cR8KlV3N64C3fKRHIcFFmbls2A$> > <https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$> > | 3 ++- > > > 1 file changed, 2 insertions(+), 1 deletion(-) > > > > > > diff --git a/meta/recipes-devtools/opkg/opkg_0.6.0.bb <https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!t0JG8eQqpsPs8clofb2phwJc6ONCKBjy-iQukKenLRVzBQvJrKTp8VsgKBI7cR8KlV3N64C3fKRHIcFFmbls2A$> > <https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$> > b/meta/recipes-devtools/opkg/opkg_0.6.0.bb <https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!t0JG8eQqpsPs8clofb2phwJc6ONCKBjy-iQukKenLRVzBQvJrKTp8VsgKBI7cR8KlV3N64C3fKRHIcFFmbls2A$> > <https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$> > > > index 7b351e8123..e38d9d6f3f 100644 > > > --- a/meta/recipes-devtools/opkg/opkg_0.6.0.bb <https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!t0JG8eQqpsPs8clofb2phwJc6ONCKBjy-iQukKenLRVzBQvJrKTp8VsgKBI7cR8KlV3N64C3fKRHIcFFmbls2A$> > <https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$> > > > +++ b/meta/recipes-devtools/opkg/opkg_0.6.0.bb <https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!t0JG8eQqpsPs8clofb2phwJc6ONCKBjy-iQukKenLRVzBQvJrKTp8VsgKBI7cR8KlV3N64C3fKRHIcFFmbls2A$> > <https://urldefense.com/v3/__http://opkg_0.6.0.bb__;!!FbZ0ZwI3Qg!u0KgIth5yNSO1tok49C5_xo0QPHIFLa4kZBl3vVZwB-2Ui1uYBW7nt85fKXXM_VAHZ2LdFx_-65pBFo$> > > > @@ -30,7 +30,7 @@ inherit autotools pkgconfig ptest > > > target_localstatedir := "${localstatedir}" > > > OPKGLIBDIR ??= "${target_localstatedir}/lib" > > > > > > -PACKAGECONFIG ??= "libsolv" > > > +PACKAGECONFIG ??= "libsolv zstd" > > > > Building in zstd support by default is a little suspect to me. > > > > Unless I'm mistaken, OE-core will only build xz-compressed IPKs by > > default. So zstd support would be unnecessary for a distro > integrator > > who just uses upstream OE-core. > > > > For distros which use zstd compression in their packages, I think it > > would be more appropriate to overwrite the opkg PACKAGECONFIG in a > > .bbappend. > > > > This is perhaps fine. I do wonder if there is some performance > comparison data between xz and zstd compressed ipks > with opkg, it might help users on making this choice and also if we > should consider using > zstd by default at some point or not. > > > Is there something I'm not considering here? > > > > > > > > PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ > > > gnupg gpgme libgpg-error,\ > > > @@ -39,6 +39,7 @@ PACKAGECONFIG[gpg] = > "--enable-gpg,--disable-gpg,\ > > > PACKAGECONFIG[curl] = "--enable-curl,--disable-curl,curl" > > > PACKAGECONFIG[ssl-curl] = > "--enable-ssl-curl,--disable-ssl-curl,curl openssl" > > > PACKAGECONFIG[sha256] = "--enable-sha256,--disable-sha256" > > > +PACKAGECONFIG[zstd] = "--enable-zstd,--disable-zstd,zstd" > > > PACKAGECONFIG[libsolv] = > "--with-libsolv,--without-libsolv,libsolv" > > > > > > EXTRA_OECONF:class-native = > "--localstatedir=/${@os.path.relpath('${localstatedir}', > '${STAGING_DIR_NATIVE}')} > --sysconfdir=/${@os.path.relpath('${sysconfdir}', > '${STAGING_DIR_NATIVE}')}" > > > > > > > > > > > > > -- > > Alex Stewart > > Software Engineer - NI Real-Time OS > > NI (National Instruments) > > > > [email protected] > > > > > > -=-=-=-=-=-=-=-=-=-=-=- > > Links: You receive all messages sent to this group. > > View/Reply Online (#170608): > https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_message_170608&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=_VWADuvDS3QdSEYWTnh8wNH0lOxVHty18QZynvOtpcU&e= <https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_message_170608&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=_VWADuvDS3QdSEYWTnh8wNH0lOxVHty18QZynvOtpcU&e=> > <https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_message_170608&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=_VWADuvDS3QdSEYWTnh8wNH0lOxVHty18QZynvOtpcU&e= <https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_message_170608&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=_VWADuvDS3QdSEYWTnh8wNH0lOxVHty18QZynvOtpcU&e=>> > > > Mute This Topic: > https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_mt_93654146_1997914&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=QncdL2I_pCDsz7AMbsTgStzC7bt5V2x_E_gIblrtjZY&e= <https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_mt_93654146_1997914&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=QncdL2I_pCDsz7AMbsTgStzC7bt5V2x_E_gIblrtjZY&e=> > <https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_mt_93654146_1997914&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=QncdL2I_pCDsz7AMbsTgStzC7bt5V2x_E_gIblrtjZY&e= <https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_mt_93654146_1997914&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=QncdL2I_pCDsz7AMbsTgStzC7bt5V2x_E_gIblrtjZY&e=>> > > > Group Owner: [email protected] <mailto:openembedded-core%[email protected]> > <mailto:openembedded-core%[email protected] <mailto:openembedded-core%[email protected]>> > > Unsubscribe: > https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_unsub&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=PxAE_3yGzbSZ21-LHz_1r3gd2_STPoZcrYDeu9vd_GM&e= <https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_unsub&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=PxAE_3yGzbSZ21-LHz_1r3gd2_STPoZcrYDeu9vd_GM&e=> > <https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_unsub&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=PxAE_3yGzbSZ21-LHz_1r3gd2_STPoZcrYDeu9vd_GM&e= <https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_openembedded-2Dcore_unsub&d=DwIBaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=ixV5wgZimxpcSCog35wonZKeFaeEVKUbESWNj_K-1j0&m=KvCRyrKojziBs0h9aEjEfI4R_J4FnZIK8nDBqKo_I1CpnrIJ78DBxgMrvY25Gwyh&s=PxAE_3yGzbSZ21-LHz_1r3gd2_STPoZcrYDeu9vd_GM&e=>> > [[email protected]] > > -=-=-=-=-=-=-=-=-=-=-=- > >Quoted 9 lines
> -- Alex Stewart Software Engineer - NI Real-Time OS NI (National Instruments) [email protected]--
Alex Stewart
Software Engineer - NI Real-Time OS
NI (National Instruments)
I tested it a bit more today. I used the standard poky's local.conf and added those lines. You can see that at zstd level 9 there is still a significant difference in the compression ratio with xz for gcc-dbg which is a big file. At zstd level 19, gcc-dbg is 241MB big instead of 214MB with xz (12.6% more).
PACKAGE_CLASSES = "package_ipk"ZSTD_DEFAULTS = "-T0 -19" PACKAGECONFIG:append:pn-opkg-native = " zstd" OPKGBUILDCMD = "opkg-build -Z zstd -a ${ZSTD_DEFAULTS}"
Zstd at level 9:
build$ ls -lh tmp-zstd-level9/deploy/ipk/core2-64/ --sort=size | head -n10 total 1.5G
-rw-r--r-- 3 ecordonnier ecordonnier 302M Sep 28 18:33 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 249M Sep 28 18:33 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 106M Sep 28 18:33 openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 44M Sep 28 18:33
binutils-dbg_2.39-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 26M Sep 28 18:33
perl-ptest_5.36.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 23M Sep 28 18:33
gcc_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 15M Sep 28 18:32
elfutils-ptest_0.187-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 15M Sep 28 18:33
gcc-src_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 13M Sep 28 18:33
coreutils-dbg_9.1-r0_core2-64.ipk
zstd at level 19:
build$ ls -lh tmp-zstd-level19/deploy/ipk/core2-64/ --sort=size | head -n10 total 1.1G
-rw-r--r-- 3 ecordonnier ecordonnier 241M Sep 28 18:04 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 213M Sep 28 18:03 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 36M Sep 28 17:58
binutils-dbg_2.39-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 30M Sep 28 18:00
openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 22M Sep 28 17:59
perl-ptest_5.36.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 20M Sep 28 17:58
gcc_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 14M Sep 28 17:57
elfutils-ptest_0.187-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 12M Sep 28 17:58
gcc-src_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 12M Sep 28 17:57
g++_12.2.0-r0_core2-64.ipk
xz at level 9 (the default):
build$ ls -lh tmp-xz/deploy/ipk/core2-64/ --sort=size | head -n10 total 963M
-rw-r--r-- 3 ecordonnier ecordonnier 214M Sep 14 10:44 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 193M Sep 14 10:44 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 35M Sep 14 10:44
openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 32M Sep 14 10:44
binutils-dbg_2.39-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 20M Sep 14 10:44
perl-ptest_5.36.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 19M Sep 14 10:44
gcc_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 13M Sep 14 10:44
elfutils-ptest_0.187-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 12M Sep 14 10:44
gcc-src_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 11M Sep 14 10:44
g++_12.2.0-r0_core2-64.ipk
Quoted 378 lines · Khem Raj
Quoted 7 lines · Alex Stewart
Thanks for checking.
I'd be interested to know if setting a higher compression level for zstd can get us to a similar compression ratio to xz. If so, then I think it could be some real value to distro maintainers to be able to *tune* their compression.
yeah it will be interesting to say try something like level 9 but I think times
might regress with that but it might be good to know the balance and perhaps
suggest size mode and performance mode of zstd instead of xzQuoted 71 lines
That's not blocking for your new PR though.
Quoted 66 lines · Etienne Cordonnier
Also note that zstd's default compression level is 3 per default (from a 1 to 19 scale). I did not test other compression levels.
On Wed, Sep 14, 2022 at 11:58 AM Etienne Cordonnier <[email protected]> wrote:
I ran a build of core-image-full-cmdline using xz and zstd, using pre-populated downloads and sstate-cache directories but with empty tmp directory. Here are the numbers:
zstd:
bitbake core-image-full-cmdline took 2m52.768s (real), the resulting directory tmp/deploy/ipk is 1.6GB big.
xz:
bitbake core-image-full-cmdline took 4m14.214s (real), the resulting directory tmp/deploy/ipk/ is 996M bigSo the build with xz is 47% slower (254/172) than with zstd for this "incremental build" use-case.
See the 5 biggest packages, the difference in compression-ratio increases with big files:
build$ ls -lh tmp-zstd/deploy/ipk/core2-64/ --sort=size | head -n5 total 1.6G
-rw-r--r-- 3 ecordonnier ecordonnier 331M Sep 14 10:39 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 264M Sep 14 10:39 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 113M Sep 14 10:39 openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 47M Sep 14 10:39
binutils-dbg_2.39-r0_core2-64.ipk
build$ ls -lh tmp-xz/deploy/ipk/core2-64/ --sort=size | head -n5 total 963M
-rw-r--r-- 3 ecordonnier ecordonnier 214M Sep 14 10:44 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 193M Sep 14 10:44 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 35M Sep 14 10:44
openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 32M Sep 14 10:44
binutils-dbg_2.39-r0_core2-64.ipk
...I think for use-cases where the size of the ipk packages matters, xz is better. For use-cases where it does not matter (ipk packages not deployed), build-time matters more than compression-ratio and zstd is better.
Regarding the enablement of zstd in opkg per default, I'll send a new version of the patch without this line.
My thinking for enabling zstd per default in opkg was that zstd is already enabled per default in libarchive's PACKAGECONFIG, and disabling zstd in opkg's PACKAGECONFIG removes only a few lines of code from opkg (opkg uses libarchive for doing the actual compression/decompression).On Tue, Sep 13, 2022 at 11:57 PM Alex Stewart
<[email protected]> wrote:On 9/13/22 15:20, Alex Feinman wrote:
> I do have some numbers. When I was selling this change internally, I
> did a comparison on our internal build.
> Combined write IPK times (Σ t
do_package_write_ipk)Quoted 46 lines
Quoted 46 lines
> xz 162m 35s > gz 52m 13s > zstd 33m 49s> Compression rate for zstd was closer to xz than to gz but not as good
> as xz. For systems that have to cache packages on the device with
> limited storage xz might be a better option, but for the bulk of
> projects zstd is the best choice
> Additionally, zstd offers much faster decompression than xz so the
> rootfs build step that includes unpacking all of the ipks, takes 3m
> 58s with xz and 2m 44s with zstd.
> One other thing of note - if your build includes debug packages, some
> may be quite large. E.g. one of our components produces a 2.2 GB debug
> package (uncompressed). On large files xz requires a disproportionally
> large amount of time resulting in 15 minutes needed to simply write
> ipk for the abovementioned packages, whereas zstd took about 45 sec.
> For frequent tasks like bitbaking a single package this translates in
> a lot of saved time.Those are certainly compelling performance improvements. Assuming that
the final data-segment size is within 5%-ish of xz, then I would agree
with the rest of the thread that it should probably be the contemporary
default.And if we make it the default compressor for OE IPKs, then obviously my
criticism in the original PR is satisfied.> Bottom line - I think making xz a default package compressor was not
> entirely thought through. gzip or zstd is what the default should be.Quoted 6 lines
Quoted 6 lines
ZStandard support was only added to opkg last September [1]. Before
that, xz was the new hotness that replaced gzip. :)[1]
Quoted 1 line
Quoted 1 line
<Quoted 14 lines
Quoted 14 lines
> One final note: I could not find a reasonable explanation for why
> opkg-tools require code changes to support a different compressor. BSD
> tar and GNU tar both can easily accept compressors that they have no
> idea about (via -I option) because all of them provide a unified
> command line interface for use in pipes. If this were done similar to
> tar, we could have used any compressor we wanted, includingthe
Quoted 12 lines
Quoted 12 lines
> multithreaded versions (zstdmt)
Well, presumably IPK creation tools can only support the matrix of
compression algorithms which your opkg binary can decompress. I suppose
someone could try to implement a plugable compression module system for
opkg. But given that nearly everyone uses opkg in an embedded context,
I'm not sure it would get much use.Quoted 12 lines
Quoted 4 lines
> > On Tue, Sep 13, 2022 at 12:43 PM Khem Raj <[email protected]> wrote: >Quoted 8 lines
> On Tue, Sep 13, 2022 at 12:19 PM Alex Stewart > <[email protected]> wrote: > > > > ACK from me - apart from enabling zstd by default. > > > > On 9/13/22 07:37, Etienne Cordonnier via lists.openembedded.org <Quoted 3 lines
Quoted 3 lines
> <Quoted 3 lines
Quoted 3 lines
> wrote:
> > > This allows the use of zstd for opkg packages byusing
Quoted 5 lines
Quoted 5 lines
> OPKGBUILDCMD: > > > OPKGBUILDCMD = "opkg-build -Z zstd" > > > > > > Signed-off-by: Alex Feinman <[email protected]> > > > Signed-off-by: Etienne Cordonnier <Quoted 3 lines
Quoted 3 lines
> > > --- > > > meta/recipes-devtools/opkg/opkg_0.6.0.bb <Quoted 3 lines
Quoted 3 lines
> <Quoted 7 lines
Quoted 7 lines
> | 3 ++- > > > 1 file changed, 2 insertions(+), 1 deletion(-) > > > > > > diff --git a/meta/recipes-devtools/opkg/opkg_0.6.0.bb <Quoted 3 lines
Quoted 3 lines
> <Quoted 3 lines
Quoted 3 lines
> b/meta/recipes-devtools/opkg/opkg_0.6.0.bb <Quoted 3 lines
Quoted 3 lines
> <Quoted 4 lines
Quoted 4 lines
> > > index 7b351e8123..e38d9d6f3f 100644 > > > --- a/meta/recipes-devtools/opkg/opkg_0.6.0.bb <Quoted 3 lines
Quoted 3 lines
> <Quoted 3 lines
Quoted 3 lines
> > > +++ b/meta/recipes-devtools/opkg/opkg_0.6.0.bb <Quoted 3 lines
Quoted 3 lines
> <Quoted 14 lines
Quoted 9 lines
> > > @@ -30,7 +30,7 @@ inherit autotools pkgconfig ptest > > > target_localstatedir := "${localstatedir}" > > > OPKGLIBDIR ??= "${target_localstatedir}/lib" > > > > > > -PACKAGECONFIG ??= "libsolv" > > > +PACKAGECONFIG ??= "libsolv zstd" > > > > Building in zstd support by default is a littleQuoted 5 lines
suspect to me.
> >
> > Unless I'm mistaken, OE-core will only build
xz-compressed IPKs by
> > default. So zstd support would be unnecessary for adistro
Quoted 50 lines
Quoted 50 lines
> integrator > > who just uses upstream OE-core. > > > > For distros which use zstd compression in their packages, I think it > > would be more appropriate to overwrite the opkg PACKAGECONFIG in a > > .bbappend. > > > > This is perhaps fine. I do wonder if there is some performance > comparison data between xz and zstd compressed ipks > with opkg, it might help users on making this choice and also if we > should consider using > zstd by default at some point or not. > > > Is there something I'm not considering here? > > > > > > > > PACKAGECONFIG[gpg] = "--enable-gpg,--disable-gpg,\ > > > gnupg gpgme libgpg-error,\ > > > @@ -39,6 +39,7 @@ PACKAGECONFIG[gpg] = > "--enable-gpg,--disable-gpg,\ > > > PACKAGECONFIG[curl] = "--enable-curl,--disable-curl,curl" > > > PACKAGECONFIG[ssl-curl] = > "--enable-ssl-curl,--disable-ssl-curl,curl openssl" > > > PACKAGECONFIG[sha256] = "--enable-sha256,--disable-sha256" > > > +PACKAGECONFIG[zstd] = "--enable-zstd,--disable-zstd,zstd" > > > PACKAGECONFIG[libsolv] = > "--with-libsolv,--without-libsolv,libsolv" > > > > > > EXTRA_OECONF:class-native = > "--localstatedir=/${@os.path.relpath('${localstatedir}', > '${STAGING_DIR_NATIVE}')} > --sysconfdir=/${@os.path.relpath('${sysconfdir}', > '${STAGING_DIR_NATIVE}')}" > > > > > > > > > > > > > -- > > Alex Stewart > > Software Engineer - NI Real-Time OS > > NI (National Instruments) > >Quoted 8 lines
Quoted 8 lines
> > [email protected] > > > > > > -=-=-=-=-=-=-=-=-=-=-=- > > Links: You receive all messages sent to this group. > > View/Reply Online (#170608): >Quoted 1 line
Quoted 1 line
<Quoted 3 lines
Quoted 3 lines
> <Quoted 1 line
Quoted 1 line
<Quoted 5 lines
Quoted 5 lines
> > > Mute This Topic: >Quoted 1 line
Quoted 1 line
<Quoted 3 lines
Quoted 3 lines
> <Quoted 1 line
Quoted 1 line
<Quoted 10 lines
Quoted 10 lines
>
> > Group Owner:
[email protected] <mailto:openembedded-core%[email protected]>
> <mailto:openembedded-core%[email protected]
<mailto:openembedded-core%[email protected]>>
> > Unsubscribe:
>Quoted 1 line
Quoted 1 line
<Quoted 3 lines
Quoted 3 lines
> <Quoted 1 line
Quoted 1 line
<Quoted 21 lines
Quoted 13 lines
> [[email protected]] > > -=-=-=-=-=-=-=-=-=-=-=- > > > -- Alex Stewart Software Engineer - NI Real-Time OS NI (National Instruments) [email protected]--
Alex Stewart
Software Engineer - NI Real-Time OS
NI (National Instruments)
Thanks for doing the research.
What was the build time on the zstd L19 compression vs. xz? Was there still an improvement?
Quoted 97 lines · Etienne Cordonnier
I tested it a bit more today. I used the standard poky's local.conf and added those lines. You can see that at zstd level 9 there is still a significant difference in the compression ratio with xz for gcc-dbg which is a big file. At zstd level 19, gcc-dbg is 241MB big instead of 214MB with xz (12.6% more).
PACKAGE_CLASSES= "package_ipk"ZSTD_DEFAULTS= "-T0 -19" PACKAGECONFIG:append:pn-opkg-native = " zstd" OPKGBUILDCMD = "opkg-build -Z zstd -a ${ZSTD_DEFAULTS}"Zstd at level 9:
build$ ls -lh tmp-zstd-level9/deploy/ipk/core2-64/ --sort=size | head -n10
total 1.5G
-rw-r--r-- 3 ecordonnier ecordonnier 302M Sep 28 18:33 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 249M Sep 28 18:33 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 106M Sep 28 18:33 openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 44M Sep 28 18:33 binutils-dbg_2.39-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 26M Sep 28 18:33 perl-ptest_5.36.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 23M Sep 28 18:33 gcc_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 15M Sep 28 18:32 elfutils-ptest_0.187-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 15M Sep 28 18:33 gcc-src_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 13M Sep 28 18:33 coreutils-dbg_9.1-r0_core2-64.ipkzstd at level 19:
build$ ls -lh tmp-zstd-level19/deploy/ipk/core2-64/ --sort=size | head -n10 total 1.1G
-rw-r--r-- 3 ecordonnier ecordonnier 241M Sep 28 18:04 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 213M Sep 28 18:03 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 36M Sep 28 17:58 binutils-dbg_2.39-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 30M Sep 28 18:00 openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 22M Sep 28 17:59 perl-ptest_5.36.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 20M Sep 28 17:58 gcc_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 14M Sep 28 17:57 elfutils-ptest_0.187-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 12M Sep 28 17:58 gcc-src_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 12M Sep 28 17:57 g++_12.2.0-r0_core2-64.ipkxz at level 9 (the default):
build$ ls -lh tmp-xz/deploy/ipk/core2-64/ --sort=size | head -n10 total 963M
-rw-r--r-- 3 ecordonnier ecordonnier 214M Sep 14 10:44 gcc-dbg_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 193M Sep 14 10:44 openssl-dbg_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 35M Sep 14 10:44 openssl-ptest_3.0.5-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 32M Sep 14 10:44 binutils-dbg_2.39-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 20M Sep 14 10:44 perl-ptest_5.36.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 19M Sep 14 10:44 gcc_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 13M Sep 14 10:44 elfutils-ptest_0.187-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 12M Sep 14 10:44 gcc-src_12.2.0-r0_core2-64.ipk
-rw-r--r-- 3 ecordonnier ecordonnier 11M Sep 14 10:44 g++_12.2.0-r0_core2-64.ipkOn Wed, Sep 14, 2022 at 5:42 PM Khem Raj <[email protected]> wrote:
On Wed, Sep 14, 2022 at 8:37 AM Alex Stewart <[email protected]> wrote:
>
> Thanks for checking.
>
> I'd be interested to know if setting a higher compression level for zstd
> can get us to a similar compression ratio to xz. If so, then I think it
> could be some real value to distro maintainers to be able to *tune* > their compression.yeah it will be interesting to say try something like level 9 but I think times
might regress with that but it might be good to know the balance and perhaps
suggest size mode and performance mode of zstd instead of xz
Signature · 5 lines
Alex Stewart Software Engineer - NI Real-Time OS NI (National Instruments) [email protected]