Skip to contentSkip to conversation

gzip: upgrade 1.13 -> 1.14

PATCH In review Lei Maohui 5 participants 9 messages Apr 17, 2025 – Apr 27, 2025

lore
Share & tools
lore

Lei Maohui

author
#0 lore raw

PATCH 1/1 gzip: upgrade 1.13 -> 1.14

fix issue:
|ERROR: gzip-1.14-r0 do_package_qa: QA Issue: File /usr/lib/gzip/ptest/src/tests/Makefile in package gzip-ptest contains reference to TMPDIR ildpaths] |ERROR: gzip-1.14-r0 do_package_qa: Fatal QA errors were found, failing task.

error point in /usr/lib/gzip/ptest/src/tests/Makefile of gzip-ptest:

|BUILD_LDFLAGS = -L/gzip/1.14/recipe-sysroot-native/usr/lib                         -L/gzip/1.14/recipe-sysroot-native/lib
|      -Wl,--enable-new-dtags                         -Wl,-rpath-link,/gzip/1.14/recipe-sysroot-native/usr/lib                         -Wl,-r
|path-link,/gzip/1.14/recipe-sysroot-native/lib                         -Wl,-rpath,/gzip/1.14/recipe-sysroot-native/usr/lib
|       -Wl,-rpath,/gzip/1.14/recipe-sysroot-native/lib                         -Wl,-O1 -Wl,--allow-shlib-undefined -Wl,--dynamic-linker=/mnt
|/test/build_auh/tmp/sysroots-uninative/x86_64-linux/lib/ld-linux-x86-64.so.2 -pthread

ptest pass after delete BUILD_LDFLAGS in Makefile

ptest result:

|============================================================================
|Testsuite summary for gzip 1.14
|============================================================================
|# TOTAL: 30
|# PASS:  28
|# SKIP:  2
|# XFAIL: 0
|# FAIL:  0
|# XPASS: 0
|# ERROR: 0

License-Update: copyright years update and simplifies compliance by replacing physical contact instructions with a permanent web reference

---
 ...rence-fix-invalid-option-of-od-issue.patch | 36 +++++++++++++++++++
 .../wrong-path-fix.patch                      | 13 ++++---
 .../gzip/{gzip_1.13.bb => gzip_1.14.bb}       |  7 ++--
 3 files changed, 47 insertions(+), 9 deletions(-)
 create mode 100644 meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch
 rename meta/recipes-extended/gzip/{gzip-1.13 => gzip-1.14}/wrong-path-fix.patch (76%)
 rename meta/recipes-extended/gzip/{gzip_1.13.bb => gzip_1.14.bb} (80%)
meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch new file +36 -0
new file mode 100644
index 0000000000..cc7ddbef5a
@@ -0,0 +1,36 @@
+From fe8f7d75ca16e8c1c53b210dcdf130b91b57c252 Mon Sep 17 00:00:00 2001
+From: Lei Maohui <[email protected]>
+Date: Thu, 17 Apr 2025 01:15:06 +0000
+Subject: [PATCH] tests/reference: fix invalid option of od issue
+
+ptest issue:
+| od: invalid option -- 'A'
+| BusyBox v1.37.0 () multi-call binary.
+
+replace "od" with "perl" according to the comments before error point
+|# Ensure that compressing these simple strings always produces the same bytes.
+|# If using "od" is not portable enough, consider using this:
+|# perl -ne 'printf "%02x ", ord($_) for split //'
+|for i in '' a b c yyy zzzzzzzzzzz; do
+|  echo $i: $(printf %s "$i" | gzip | od -An -tx1 | tr -d '\n')
+|done
+
+Upstream-Status: Inappropriate [oe-core specific]
+Signed-off-by: Lei Maohui <[email protected]>
+---
+ tests/reference | 2 +-
+ 1 file changed, 1 insertion(+), 1 deletion(-)
+
+diff --git a/tests/reference b/tests/reference
+index add962e..bfd6ad3 100755
+--- a/tests/reference
++++ b/tests/reference
+@@ -43,7 +43,7 @@ EOF
+ # If using "od" is not portable enough, consider using this:
+ # perl -ne 'printf "%02x ", ord($_) for split //'
+ for i in '' a b c yyy zzzzzzzzzzz; do
+-  echo $i: $(printf %s "$i" | gzip | od -An -tx1 | tr -d '\n')
++  echo $i: $(printf %s "$i" | gzip | perl -ne 'printf "%02x ", ord($_) for split //')
+ done > out || framework_failure_
+ 
+ compare exp out || fail=1
meta/recipes-extended/gzip/gzip-1.14/wrong-path-fix.patch renamed from meta/recipes-extended/gzip/gzip-1.13/wrong-path-fix.patch +6 -7
similarity index 76%
rename from meta/recipes-extended/gzip/gzip-1.13/wrong-path-fix.patch
rename to meta/recipes-extended/gzip/gzip-1.14/wrong-path-fix.patch
index 7f9e249de8..4d5e7a8e02 100644
@@ -1,4 +1,7 @@
-fix MakeMaker issues with using wrong SHELL/GREP
+From eda9b1d08c517acbdc5b26c24c94a3985f29c749 Mon Sep 17 00:00:00 2001
+From: Ming Liu <[email protected]>
+Date: Fri, 21 Nov 2014 04:50:57 -0500
+Subject: [PATCH] fix MakeMaker issues with using wrong SHELL/GREP
 
 A set of substitution is being processed to all target scripts with sed by
 replacing some key words with the detected values at configure time, this
@@ -12,16 +15,15 @@ instead.
 Signed-off-by: Ming Liu <[email protected]>
 
 Upstream-Status: Pending
-
 ---
  Makefile.am | 3 +--
  1 file changed, 1 insertion(+), 2 deletions(-)
 
 diff --git a/Makefile.am b/Makefile.am
-index 4f51b61..80a5ddf 100644
+index 23e0d3e..3455878 100644
 --- a/Makefile.am
 +++ b/Makefile.am
-@@ -93,8 +93,7 @@ SUFFIXES = .in
+@@ -95,8 +95,7 @@ SUFFIXES = .in
  .in:
  	$(AM_V_GEN)rm -f $@-t $@ \
  	  && sed \
@@ -31,6 +33,3 @@ index 4f51b61..80a5ddf 100644
  		-e "s|'gzip'|$(GZIP_TRANSFORMED)|g" \
  		-e "s|'zdiff'|$(ZDIFF_TRANSFORMED)|g" \
  		-e "s|'zgrep'|$(ZGREP_TRANSFORMED)|g" \
---
-2.7.4
-
meta/recipes-extended/gzip/gzip_1.14.bb renamed from meta/recipes-extended/gzip/gzip_1.13.bb +5 -2
similarity index 80%
rename from meta/recipes-extended/gzip/gzip_1.13.bb
rename to meta/recipes-extended/gzip/gzip_1.14.bb
index fd846b30a5..f12dcb1e77 100644
@@ -6,11 +6,13 @@ LICENSE = "GPL-3.0-or-later"
 
 SRC_URI = "${GNU_MIRROR}/gzip/${BP}.tar.gz \
            file://run-ptest \
+           file://0001-tests-reference-fix-invalid-option-of-od-issue.patch \
            "
 SRC_URI:append:class-target = " file://wrong-path-fix.patch"
 
 LIC_FILES_CHKSUM = "file://COPYING;md5=1ebbd3e34237af26da5dc08a4e440464 \
-                    file://gzip.h;beginline=8;endline=20;md5=6e47caaa630e0c8bf9f1bc8d94a8ed0e"
+                    file://gzip.h;beginline=8;endline=20;md5=a22158dc3dd3f5cf6e5a556940a49212 \
+                    "
 
 PROVIDES:append:class-native = " gzip-replacement-native"
 
@@ -33,9 +35,10 @@ do_install_ptest() {
             -e 's/^Makefile: ..*/Makefile: /'           \
             -e 's,--sysroot=${STAGING_DIR_TARGET},,g'   \
             -e 's|${DEBUG_PREFIX_MAP}||g' \
+            -e 's|${BUILD_LDFLAGS}||g' \
             -e 's:${HOSTTOOLS_DIR}/::g'                 \
             -e 's:${BASE_WORKDIR}/${MULTIMACH_TARGET_SYS}::g' \
             ${B}/tests/Makefile > ${D}${PTEST_PATH}/src/tests/Makefile
 }
 
-SRC_URI[sha256sum] = "20fc818aeebae87cdbf209d35141ad9d3cf312b35a5e6be61bfcfbf9eddd212a"
+SRC_URI[sha256sum] = "613d6ea44f1248d7370c7ccdeee0dd0017a09e6c39de894b3c6f03f981191c6b"
Signature · 1 line
2.43.0
#1 lore raw

meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch line 8

Quoted 5 lines · leimaohui via lists.openembedded.org
meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch new file +5 -0
+Subject: [PATCH] tests/reference: fix invalid option of od issue
+
+ptest issue:
+| od: invalid option -- 'A'
+| BusyBox v1.37.0 () multi-call binary.

Instead of working around busybox od limitations with a patch, ptest package should rdepend on coreutils to pull in the real original od.

Alex

#2 lore raw
Quoted 10 lines · Alexander Kanavin via lists.openembedded.org

meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch line 8

Quoted 5 lines · leimaohui via lists.openembedded.org
meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch new file +5 -0
+Subject: [PATCH] tests/reference: fix invalid option of od issue
+
+ptest issue:
+| od: invalid option -- 'A'
+| BusyBox v1.37.0 () multi-call binary.

Instead of working around busybox od limitations with a patch, ptest package should rdepend on coreutils to pull in the real original od.

Agreed, I've been meaning to reply to say exactly this!

Cheers,

Richard

#3 lore raw
Quoted 13 lines · Richard Purdie via lists.openembedded.org
Quoted 10 lines · Alexander Kanavin via lists.openembedded.org

meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch line 8

Quoted 5 lines · leimaohui via lists.openembedded.org
meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch new file +5 -0
+Subject: [PATCH] tests/reference: fix invalid option of od issue
+
+ptest issue:
+| od: invalid option -- 'A'
+| BusyBox v1.37.0 () multi-call binary.

Instead of working around busybox od limitations with a patch, ptest package should rdepend on coreutils to pull in the real original od.

Quoted 2 lines

Agreed, I've been meaning to reply to say exactly this!

Consider other side of this as well.
Many embedded systems do use busybox heavily, imposing dependencies for tests like this would mean ptests are less relevant for such usecases since bringing coreutils might also shunt other busybox utilities along the way.

I would ask this patch to be proposed to gzip upstream first though.

Quoted 14 lines

Cheers,

Richard

-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#215233): https://lists.openembedded.org/g/openembedded-core/message/215233
Mute This Topic: https://lists.openembedded.org/mt/112310990/1997914
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

#4 lore raw
Quoted 7 lines · Khem Raj

Consider other side of this as well.
Many embedded systems do use busybox heavily, imposing dependencies for tests like this would mean ptests are less relevant for such usecases since bringing coreutils might also shunt other busybox utilities along the way.

I would ask this patch to be proposed to gzip upstream first though.

Nearly all upstreams run these tests with a GNU userspace environment on some classic binary distro. It's not that we impose GNU on ptests, it's that we don't want to deviate from upstream expectations, as doing that only results in upstream unknowingly introducing regressions and us fixing them after the fact, with no guarantee that upstreams will take the fixes. Version updates are hard enough as it is.

I can kinda see the point in fixing ptests to work with musl, but fixing them to work with busybox is a step too far. There's a lot of ptests that already pull in various GNU pieces, so adding gzip to that set will not make it any worse.

Alex

#5 lore raw
Quoted 9 lines · Alexander Kanavin
Quoted 7 lines · Khem Raj

Consider other side of this as well.
Many embedded systems do use busybox heavily, imposing dependencies for tests like this would mean ptests are less relevant for such usecases since bringing coreutils might also shunt other busybox utilities along the way.

I would ask this patch to be proposed to gzip upstream first though.

Quoted 7 lines

Nearly all upstreams run these tests with a GNU userspace environment on some classic binary distro. It's not that we impose GNU on ptests, it's that we don't want to deviate from upstream expectations, as doing that only results in upstream unknowingly introducing regressions and us fixing them after the fact, with no guarantee that upstreams will take the fixes. Version updates are hard enough as it is.

That is fine, and I can empathize. OEs usecases are unlike classic binary distros, otherwise thats what users would use. We need to be mindful of end users too. A bit of extra portability helps us, if we can get upstreams to support that then it helps our usecase.

Quoted 7 lines

I can kinda see the point in fixing ptests to work with musl, but fixing them to work with busybox is a step too far. There's a lot of ptests that already pull in various GNU pieces, so adding gzip to that set will not make it any worse.

Alex

#6 lore raw
In reply to Alexander Kanavin
Quoted 14 lines

-----邮件原件-----
发件人: [email protected] <[email protected]> 代表 Alexander Kanavin via lists.openembedded.org
发送时间: 2025年4月22日 23:25
收件人: Lei, Maohui/雷 茂慧 <[email protected]>
抄送: Denys Dmytriyenko <[email protected]>; [email protected]
主题: Re: [OE-core] [PATCH] gzip: upgrade 1.13 -> 1.14

On Thu, 17 Apr 2025 at 12:21, leimaohui via lists.openembedded.org <[email protected]> wrote:

Quoted 8 lines

meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch line 8

Quoted 5 lines
meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch new file +5 -0
+Subject: [PATCH] tests/reference: fix invalid option of od issue
+
+ptest issue:
+| od: invalid option -- 'A'
+| BusyBox v1.37.0 () multi-call binary.

Instead of working around busybox od limitations with a patch, ptest package should rdepend on coreutils to pull in the real original od.

I will continue this patch on behalf of my colleague

Thank you, I know add rdepend on coreutils is the usual approach taken by oe-core for such cases. But have you noticed that gzip source code recommended to use perl instead of "od" if " If using "od" is not portable enough" right before the error point

$ vi gzip-1.14/ tests/reference
42 # Ensure that compressing these simple strings always produces the same bytes.
43 # If using "od" is not portable enough, consider using this:
44 # perl -ne 'printf "%02x ", ord($_) for split //'

Liu Yiding

Quoted 2 lines

Alex

#7 lore raw
Quoted 29 lines · Yiding Liu (Fujitsu)
Quoted 14 lines

-----邮件原件-----
发件人: [email protected] <[email protected]> 代表 Alexander Kanavin via lists.openembedded.org
发送时间: 2025年4月22日 23:25
收件人: Lei, Maohui/雷 茂慧 <[email protected]>
抄送: Denys Dmytriyenko <[email protected]>; [email protected]
主题: Re: [OE-core] [PATCH] gzip: upgrade 1.13 -> 1.14

On Thu, 17 Apr 2025 at 12:21, leimaohui via lists.openembedded.org <[email protected]> wrote:

Quoted 8 lines

meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch line 8

Quoted 5 lines
meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch new file +5 -0
+Subject: [PATCH] tests/reference: fix invalid option of od issue
+
+ptest issue:
+| od: invalid option -- 'A'
+| BusyBox v1.37.0 () multi-call binary.

Instead of working around busybox od limitations with a patch, ptest package should rdepend on coreutils to pull in the real original od.

I will continue this patch on behalf of my colleague

Thank you, I know add rdepend on coreutils is the usual approach taken by oe-core for such cases. But have you noticed that gzip source code recommended to use perl instead of "od" if " If using "od" is not portable enough" right before the error point

Quoted 4 lines
$ vi gzip-1.14/ tests/reference
42 # Ensure that compressing these simple strings always produces the same bytes.
43 # If using "od" is not portable enough, consider using this:
44 # perl -ne 'printf "%02x ", ord($_) for split //'

It is a maintainability question. Is carrying a patch doing this worth the effort of doing so? It adds maintenance overhead for us and I'm not sure we gain any significant benefits from it.

I think I'd prefer just to add a ptest package specific dependency on coreutils rather than carrying a patch we could never upstream. We should add a comment about why it is needed.

Mentioning the issue to upstream might able be helpful as they might change their test if they know it does cause downstreams some challenges.

Cheers,

Richard

#8 lore raw
Quoted 54 lines

-----Original Message-----
From: Richard Purdie <[email protected]>
Sent: 2025年4月24日 18:43
To: Liu, Yiding/刘 乙丁 <[email protected]>; [email protected]; Khem Raj <[email protected]>
Cc: Denys Dmytriyenko <[email protected]>; [email protected]
Subject: Re: 回复: [OE-core] [PATCH] gzip: upgrade 1.13 -> 1.14

Quoted 5 lines · Yiding Liu (Fujitsu)
Quoted 5 lines

-----邮件原件-----
发件人: [email protected] <[email protected]> 代表 Alexander Kanavin

via

Quoted 17 lines
Quoted 9 lines

lists.openembedded.org
发送时间: 2025年4月22日 23:25
收件人: Lei, Maohui/雷 茂慧 <[email protected]> 抄送: Denys Dmytriyenko <[email protected]>; [email protected]
主题: Re: [OE-core] [PATCH] gzip: upgrade 1.13 -> 1.14

On Thu, 17 Apr 2025 at 12:21, leimaohui via lists.openembedded.org <[email protected]> wrote:

Quoted 8 lines

meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch line 8

Quoted 5 lines
meta/recipes-extended/gzip/gzip-1.14/0001-tests-reference-fix-invalid-option-of-od-issue.patch new file +5 -0
+Subject: [PATCH] tests/reference: fix invalid option of od issue
+
+ptest issue:
+| od: invalid option -- 'A'
+| BusyBox v1.37.0 () multi-call binary.

Instead of working around busybox od limitations with a patch, ptest package should rdepend on coreutils to pull in the real original od.

Quoted 3 lines

I will continue this patch on behalf of my colleague

Thank you, I know add rdepend on coreutils is the usual approach taken by

oe-core for such cases.

Quoted 1 line

But have you noticed that gzip source code recommended to use perl instead of

"od" if " If using "od"

Quoted 5 lines

is not portable enough" right before the error point

$ vi gzip-1.14/ tests/reference
42 # Ensure that compressing these simple strings always produces the same

bytes.

Quoted 2 lines

43 # If using "od" is not portable enough, consider using this:
44 # perl -ne 'printf "%02x ", ord($_) for split //'

It is a maintainability question. Is carrying a patch doing this worth the effort of doing so? It adds maintenance overhead for us and I'm not sure we gain any significant benefits from it.

Quoted 6 lines

I think I'd prefer just to add a ptest package specific dependency on coreutils rather than carrying a patch we could never upstream. We should add a comment about why it is needed.

Mentioning the issue to upstream might able be helpful as they might change their test if they know it does cause downstreams some challenges.

Thank you
Please refer to V2 patch and I have reported to upstream that we have met this issue

Quoted 4 lines

Cheers,

Richard