Skip to contentSkip to conversation

glibc: Fix stale qemu processes left after glibc oe-selftest

PATCH In review [email protected] 4 participants 10 messages Apr 1, 2024 – Apr 5, 2024

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 glibc: Fix stale qemu processes left after glibc oe-selftest

From: Yash Shinde <[email protected]>

qemumips and qemuppc were leaving stale processes behind after running glibc oe-selftest. During analysis, it was found that it was due to "tst-scm_rights" and "tst-scm_rights-time64" tests. Disable them so that there are no stale processes left behind.

[YOCTO #15423]
https://bugzilla.yoctoproject.org/show_bug.cgi?id=15423

---
 .../glibc/glibc/0024-qemu-stale-process.patch | 45 +++++++++++++++++++
 meta/recipes-core/glibc/glibc_2.39.bb         |  1 +
 2 files changed, 46 insertions(+)
 create mode 100644 meta/recipes-core/glibc/glibc/0024-qemu-stale-process.patch
meta/recipes-core/glibc/glibc/0024-qemu-stale-process.patch new file +45 -0
new file mode 100644
index 0000000000..c28fa31b89
@@ -0,0 +1,45 @@
+glibc: Fix stale qemu processes left after glibc oe-selftest
+
+qemumips and qemuppc were leaving stale processes behind after
+running glibc oe-selftest. During analysis, it was found that
+it was due to "tst-scm_rights" and "tst-scm_rights-time64" tests.
+Disable them so that there are no stale processes left behind.
+
+[YOCTO #15423]
+https://bugzilla.yoctoproject.org/show_bug.cgi?id=15423
+
+Upstream-Status: Inappropriate [oe-core specific]
+Signed-off-by: Yash Shinde <[email protected]>
+---
+diff --git a/sysdeps/unix/sysv/linux/Makefile  b/sysdeps/unix/sysv/linux/Makefile
+--- a/sysdeps/unix/sysv/linux/Makefile    2024-03-18 01:15:49.019202881 -0700
++++ b/sysdeps/unix/sysv/linux/Makefile  2024-03-14 06:26:18.581404107 -0700
+@@ -222,7 +222,6 @@
+   tst-process_mrelease \
+   tst-quota \
+   tst-rlimit-infinity \
+-  tst-scm_rights \
+   tst-sigtimedwait \
+   tst-sync_file_range \
+   tst-sysconf-iov_max \
+@@ -233,6 +232,8 @@
+   tst-timerfd \
+   tst-ttyname-direct \
+   tst-ttyname-namespace \
++  # Skip this test to avoid stale qemu process
++  # tst-scm_rights \
+   # tests
+
+ # process_madvise requires CAP_SYS_ADMIN.
+@@ -270,9 +271,10 @@
+   tst-ntp_gettimex-time64 \
+   tst-ppoll-time64 \
+   tst-prctl-time64 \
+-  tst-scm_rights-time64 \
+   tst-sigtimedwait-time64 \
+   tst-timerfd-time64 \
++  # Skip this test to avoid stale qemu process
++  # tst-scm_rights-time64 \
+   # tests-time64
+
+ tests-clone-internal = \
meta/recipes-core/glibc/glibc_2.39.bb +1 -0
index b203060064..9122472689 100644
@@ -49,6 +49,7 @@ SRC_URI =  "${GLIBC_GIT_URI};branch=${SRCBRANCH};name=glibc \
            file://0021-fix-create-thread-failed-in-unprivileged-process-BZ-.patch \
            file://0022-Avoid-hardcoded-build-time-paths-in-the-output-binar.patch \
            file://0023-aarch64-configure-Pass-mcpu-along-with-march-to-dete.patch \
+           file://0024-qemu-stale-process.patch \
 "
 S = "${WORKDIR}/git"
 B = "${WORKDIR}/build-${TARGET_SYS}"
Signature · 1 line
2.43.0
#1 lore raw

The short log is misleading since this isn't really a fix.

You have:

glibc: Fix stale qemu processes left after glibc oe-selftest

Maybe:

glibc: Skip 2 qemu tests that can hang in oe-selftest

Skipping is what we want for now but
have you tried wrapping the hanging test in a script that runs:
strace -f -ttt -o foo <test>
?

Randy

Quoted 105 lines · Yash via lists.openembedded.org

From: Yash Shinde <[email protected]>

qemumips and qemuppc were leaving stale processes behind after running glibc oe-selftest. During analysis, it was found that it was due to "tst-scm_rights" and "tst-scm_rights-time64" tests. Disable them so that there are no stale processes left behind.

[YOCTO #15423]
https://bugzilla.yoctoproject.org/show_bug.cgi?id=15423

---
 .../glibc/glibc/0024-qemu-stale-process.patch | 45 +++++++++++++++++++
 meta/recipes-core/glibc/glibc_2.39.bb         |  1 +
 2 files changed, 46 insertions(+)
 create mode 100644

meta/recipes-core/glibc/glibc/0024-qemu-stale-process.patch

meta/recipes-core/glibc/glibc/0024-qemu-stale-process.patch new file +14 -0
new file mode 100644
index 0000000000..c28fa31b89
@@ -0,0 +1,45 @@
+glibc: Fix stale qemu processes left after glibc oe-selftest
+
+qemumips and qemuppc were leaving stale processes behind after
+running glibc oe-selftest. During analysis, it was found that
+it was due to "tst-scm_rights" and "tst-scm_rights-time64" tests.
+Disable them so that there are no stale processes left behind.
+
+[YOCTO #15423]
+https://bugzilla.yoctoproject.org/show_bug.cgi?id=15423
+
+Upstream-Status: Inappropriate [oe-core specific]
+Signed-off-by: Yash Shinde <[email protected]>
+---
+diff --git a/sysdeps/unix/sysv/linux/Makefile

b/sysdeps/unix/sysv/linux/Makefile
+--- a/sysdeps/unix/sysv/linux/Makefile 2024-03-18 01:15:49.019202881
-0700
++++ b/sysdeps/unix/sysv/linux/Makefile 2024-03-14 06:26:18.581404107 -0700
+@@ -222,7 +222,6 @@

meta/recipes-core/glibc/glibc/0024-qemu-stale-process.patch new file +28 -0
+   tst-process_mrelease \
+   tst-quota \
+   tst-rlimit-infinity \
+-  tst-scm_rights \
+   tst-sigtimedwait \
+   tst-sync_file_range \
+   tst-sysconf-iov_max \
+@@ -233,6 +232,8 @@
+   tst-timerfd \
+   tst-ttyname-direct \
+   tst-ttyname-namespace \
++  # Skip this test to avoid stale qemu process
++  # tst-scm_rights \
+   # tests
+
+ # process_madvise requires CAP_SYS_ADMIN.
+@@ -270,9 +271,10 @@
+   tst-ntp_gettimex-time64 \
+   tst-ppoll-time64 \
+   tst-prctl-time64 \
+-  tst-scm_rights-time64 \
+   tst-sigtimedwait-time64 \
+   tst-timerfd-time64 \
++  # Skip this test to avoid stale qemu process
++  # tst-scm_rights-time64 \
+   # tests-time64
+
+ tests-clone-internal = \
meta/recipes-core/glibc/glibc_2.39.bb +0 -0
index b203060064..9122472689 100644
@@ -49,6 +49,7 @@ SRC_URI =
 "${GLIBC_GIT_URI};branch=${SRCBRANCH};name=glibc \
 
 file://0021-fix-create-thread-failed-in-unprivileged-process-BZ-.patch \
 
 file://0022-Avoid-hardcoded-build-time-paths-in-the-output-binar.patch \
 

file://0023-aarch64-configure-Pass-mcpu-along-with-march-to-dete.patch \
+ file://0024-qemu-stale-process.patch \
"
S = "${WORKDIR}/git"
B = "${WORKDIR}/build-${TARGET_SYS}"
--
2.43.0

-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group. View/Reply Online (#197678):
https://lists.openembedded.org/g/openembedded-core/message/197678
Mute This Topic: https://lists.openembedded.org/mt/105259855/953399
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [ [email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

#2 lore raw
Quoted 5 lines · Randy MacLeod

**
*CAUTION: This email comes from a non Wind River email account!*
Do not click links or open attachments unless you recognize the sender and know the content is safe. The short log is misleading since this isn't really a fix.

Quoted 9 lines

You have:

glibc: Fix stale qemu processes left after glibc oe-selftest

Maybe:

glibc: Skip 2 qemu tests that can hang in oe-selftest

Will send a V2 with updated short log.

Quoted 6 lines

Skipping is what we want for now but
have you tried  wrapping the hanging test in a script that runs:
strace -f -ttt -o foo <test>
?

I tried to reproduce and strace this issue outside yocto environment for ppc/mips target but make errors were observed when cross-compiling and building. Thus, unable to capture strace logs.

With the latest poky sources, the issue is not reproducible locally and unable to check there also.

Regards,
Yash

Quoted 125 lines

Randy

On Mon, Apr 1, 2024, 04:10 Shinde, Yash via lists.openembedded.org <https://urldefense.com/v3/__http://lists.openembedded.org__;!!AjveYdw8EvQ!faU3OEL_GY3mffcv-d5kC1-J3efAL2vZ7s_BGVXjacmweHYyh4mRgvZVMhDDg7A6JW9ovUA3pSdMsSAEHYVEUA$> <[email protected]> wrote:

    From: Yash Shinde <[email protected]>

    qemumips and qemuppc were leaving stale processes behind after
    running glibc oe-selftest. During analysis, it was found that
    it was due to "tst-scm_rights" and "tst-scm_rights-time64" tests.
    Disable them so that there are no stale processes left behind.

    [YOCTO #15423]
    https://bugzilla.yoctoproject.org/show_bug.cgi?id=15423
    <https://urldefense.com/v3/__https://bugzilla.yoctoproject.org/show_bug.cgi?id=15423__;!!AjveYdw8EvQ!faU3OEL_GY3mffcv-d5kC1-J3efAL2vZ7s_BGVXjacmweHYyh4mRgvZVMhDDg7A6JW9ovUA3pSdMsSBJ3U20Jw$>

    Signed-off-by: Yash Shinde <[email protected]>
    ---
     .../glibc/glibc/0024-qemu-stale-process.patch | 45
    +++++++++++++++++++
     meta/recipes-core/glibc/glibc_2.39.bb
    <https://urldefense.com/v3/__http://glibc_2.39.bb__;!!AjveYdw8EvQ!faU3OEL_GY3mffcv-d5kC1-J3efAL2vZ7s_BGVXjacmweHYyh4mRgvZVMhDDg7A6JW9ovUA3pSdMsSD3-s9FAQ$> 
           |  1 +
     2 files changed, 46 insertions(+)
     create mode 100644
    meta/recipes-core/glibc/glibc/0024-qemu-stale-process.patch

    diff --git
    a/meta/recipes-core/glibc/glibc/0024-qemu-stale-process.patch
    b/meta/recipes-core/glibc/glibc/0024-qemu-stale-process.patch
    new file mode 100644
    index 0000000000..c28fa31b89
    --- /dev/null
    +++ b/meta/recipes-core/glibc/glibc/0024-qemu-stale-process.patch
    @@ -0,0 +1,45 @@
    +glibc: Fix stale qemu processes left after glibc oe-selftest
    +
    +qemumips and qemuppc were leaving stale processes behind after
    +running glibc oe-selftest. During analysis, it was found that
    +it was due to "tst-scm_rights" and "tst-scm_rights-time64" tests.
    +Disable them so that there are no stale processes left behind.
    +
    +[YOCTO #15423]
    +https://bugzilla.yoctoproject.org/show_bug.cgi?id=15423
    <https://urldefense.com/v3/__https://bugzilla.yoctoproject.org/show_bug.cgi?id=15423__;!!AjveYdw8EvQ!faU3OEL_GY3mffcv-d5kC1-J3efAL2vZ7s_BGVXjacmweHYyh4mRgvZVMhDDg7A6JW9ovUA3pSdMsSBJ3U20Jw$>
    +
    +Upstream-Status: Inappropriate [oe-core specific]
    +Signed-off-by: Yash Shinde <[email protected]>
    +---
    +diff --git a/sysdeps/unix/sysv/linux/Makefile
    b/sysdeps/unix/sysv/linux/Makefile
    +--- a/sysdeps/unix/sysv/linux/Makefile    2024-03-18
    01:15:49.019202881 -0700
    ++++ b/sysdeps/unix/sysv/linux/Makefile  2024-03-14
    06:26:18.581404107 -0700
    +@@ -222,7 +222,6 @@
    +   tst-process_mrelease \
    +   tst-quota \
    +   tst-rlimit-infinity \
    +-  tst-scm_rights \
    +   tst-sigtimedwait \
    +   tst-sync_file_range \
    +   tst-sysconf-iov_max \
    +@@ -233,6 +232,8 @@
    +   tst-timerfd \
    +   tst-ttyname-direct \
    +   tst-ttyname-namespace \
    ++  # Skip this test to avoid stale qemu process
    ++  # tst-scm_rights \
    +   # tests
    +
    + # process_madvise requires CAP_SYS_ADMIN.
    +@@ -270,9 +271,10 @@
    +   tst-ntp_gettimex-time64 \
    +   tst-ppoll-time64 \
    +   tst-prctl-time64 \
    +-  tst-scm_rights-time64 \
    +   tst-sigtimedwait-time64 \
    +   tst-timerfd-time64 \
    ++  # Skip this test to avoid stale qemu process
    ++  # tst-scm_rights-time64 \
    +   # tests-time64
    +
    + tests-clone-internal = \
    diff --git a/meta/recipes-core/glibc/glibc_2.39.bb
    <https://urldefense.com/v3/__http://glibc_2.39.bb__;!!AjveYdw8EvQ!faU3OEL_GY3mffcv-d5kC1-J3efAL2vZ7s_BGVXjacmweHYyh4mRgvZVMhDDg7A6JW9ovUA3pSdMsSD3-s9FAQ$>
    b/meta/recipes-core/glibc/glibc_2.39.bb
    <https://urldefense.com/v3/__http://glibc_2.39.bb__;!!AjveYdw8EvQ!faU3OEL_GY3mffcv-d5kC1-J3efAL2vZ7s_BGVXjacmweHYyh4mRgvZVMhDDg7A6JW9ovUA3pSdMsSD3-s9FAQ$>
    index b203060064..9122472689 100644
    --- a/meta/recipes-core/glibc/glibc_2.39.bb
    <https://urldefense.com/v3/__http://glibc_2.39.bb__;!!AjveYdw8EvQ!faU3OEL_GY3mffcv-d5kC1-J3efAL2vZ7s_BGVXjacmweHYyh4mRgvZVMhDDg7A6JW9ovUA3pSdMsSD3-s9FAQ$>
    +++ b/meta/recipes-core/glibc/glibc_2.39.bb
    <https://urldefense.com/v3/__http://glibc_2.39.bb__;!!AjveYdw8EvQ!faU3OEL_GY3mffcv-d5kC1-J3efAL2vZ7s_BGVXjacmweHYyh4mRgvZVMhDDg7A6JW9ovUA3pSdMsSD3-s9FAQ$>
    @@ -49,6 +49,7 @@ SRC_URI =
    "${GLIBC_GIT_URI};branch=${SRCBRANCH};name=glibc \
    file://0021-fix-create-thread-failed-in-unprivileged-process-BZ-.patch
    \
    file://0022-Avoid-hardcoded-build-time-paths-in-the-output-binar.patch
    \
    file://0023-aarch64-configure-Pass-mcpu-along-with-march-to-dete.patch
    \
    + file://0024-qemu-stale-process.patch \
     "
     S = "${WORKDIR}/git"
     B = "${WORKDIR}/build-${TARGET_SYS}"
    -- 
    2.43.0


    -=-=-=-=-=-=-=-=-=-=-=-
    Links: You receive all messages sent to this group.
    View/Reply Online (#197678):
    https://lists.openembedded.org/g/openembedded-core/message/197678
    <https://urldefense.com/v3/__https://lists.openembedded.org/g/openembedded-core/message/197678__;!!AjveYdw8EvQ!faU3OEL_GY3mffcv-d5kC1-J3efAL2vZ7s_BGVXjacmweHYyh4mRgvZVMhDDg7A6JW9ovUA3pSdMsSD_nlFatw$>
    Mute This Topic:
    https://lists.openembedded.org/mt/105259855/953399
    <https://urldefense.com/v3/__https://lists.openembedded.org/mt/105259855/953399__;!!AjveYdw8EvQ!faU3OEL_GY3mffcv-d5kC1-J3efAL2vZ7s_BGVXjacmweHYyh4mRgvZVMhDDg7A6JW9ovUA3pSdMsSDwV4R2SQ$>
    Group Owner: [email protected]
    <mailto:openembedded-core%[email protected]>
    Unsubscribe:
    https://lists.openembedded.org/g/openembedded-core/unsub
    <https://urldefense.com/v3/__https://lists.openembedded.org/g/openembedded-core/unsub__;!!AjveYdw8EvQ!faU3OEL_GY3mffcv-d5kC1-J3efAL2vZ7s_BGVXjacmweHYyh4mRgvZVMhDDg7A6JW9ovUA3pSdMsSB6aZfgkA$>
    [[email protected]]
    -=-=-=-=-=-=-=-=-=-=-=-
Signature · 2 lines
Regards,
Yash Shinde
#3 lore raw
In reply to Randy MacLeod
Quoted 6 lines · Randy MacLeod

The short log is misleading since this isn't really a fix.

You have:

glibc: Fix stale qemu processes left after glibc oe-selftest

Will send a V2 with updated short log.

Quoted 11 lines

Maybe:

glibc: Skip 2 qemu tests that can hang in oe-selftest

Skipping is what we want for now but
have you tried  wrapping the hanging test in a script that runs:
strace -f -ttt -o foo <test>
?

I tried to reproduce and strace this issue outside yocto environment for ppc/mips target but make errors were observed when cross-compiling and building. Thus, unable to capture strace logs.

With the latest poky sources, the issue is not reproducible locally and unable to check there also.

Regards,
Yash

Quoted 2 lines

Randy

#4 lore raw
In reply to [email protected]

it would be better to disable them just for architectures where it fails since it seems to be a workaround, it would be good to keep it as small as possible

Quoted 93 lines · Yash via lists.openembedded.org

From: Yash Shinde <[email protected]>

qemumips and qemuppc were leaving stale processes behind after running glibc oe-selftest. During analysis, it was found that it was due to "tst-scm_rights" and "tst-scm_rights-time64" tests. Disable them so that there are no stale processes left behind.

[YOCTO #15423]
https://bugzilla.yoctoproject.org/show_bug.cgi?id=15423

---
 .../glibc/glibc/0024-qemu-stale-process.patch | 45 +++++++++++++++++++
 meta/recipes-core/glibc/glibc_2.39.bb         |  1 +
 2 files changed, 46 insertions(+)
 create mode 100644 meta/recipes-core/glibc/glibc/0024-qemu-stale-process.patch
meta/recipes-core/glibc/glibc/0024-qemu-stale-process.patch new file +45 -0
new file mode 100644
index 0000000000..c28fa31b89
@@ -0,0 +1,45 @@
+glibc: Fix stale qemu processes left after glibc oe-selftest
+
+qemumips and qemuppc were leaving stale processes behind after
+running glibc oe-selftest. During analysis, it was found that
+it was due to "tst-scm_rights" and "tst-scm_rights-time64" tests.
+Disable them so that there are no stale processes left behind.
+
+[YOCTO #15423]
+https://bugzilla.yoctoproject.org/show_bug.cgi?id=15423
+
+Upstream-Status: Inappropriate [oe-core specific]
+Signed-off-by: Yash Shinde <[email protected]>
+---
+diff --git a/sysdeps/unix/sysv/linux/Makefile  b/sysdeps/unix/sysv/linux/Makefile
+--- a/sysdeps/unix/sysv/linux/Makefile    2024-03-18 01:15:49.019202881 -0700
++++ b/sysdeps/unix/sysv/linux/Makefile  2024-03-14 06:26:18.581404107 -0700
+@@ -222,7 +222,6 @@
+   tst-process_mrelease \
+   tst-quota \
+   tst-rlimit-infinity \
+-  tst-scm_rights \
+   tst-sigtimedwait \
+   tst-sync_file_range \
+   tst-sysconf-iov_max \
+@@ -233,6 +232,8 @@
+   tst-timerfd \
+   tst-ttyname-direct \
+   tst-ttyname-namespace \
++  # Skip this test to avoid stale qemu process
++  # tst-scm_rights \
+   # tests
+
+ # process_madvise requires CAP_SYS_ADMIN.
+@@ -270,9 +271,10 @@
+   tst-ntp_gettimex-time64 \
+   tst-ppoll-time64 \
+   tst-prctl-time64 \
+-  tst-scm_rights-time64 \
+   tst-sigtimedwait-time64 \
+   tst-timerfd-time64 \
++  # Skip this test to avoid stale qemu process
++  # tst-scm_rights-time64 \
+   # tests-time64
+
+ tests-clone-internal = \
meta/recipes-core/glibc/glibc_2.39.bb +1 -1
index b203060064..9122472689 100644
@@ -49,6 +49,7 @@ SRC_URI =  "${GLIBC_GIT_URI};branch=${SRCBRANCH};name=glibc \
            file://0021-fix-create-thread-failed-in-unprivileged-process-BZ-.patch \
            file://0022-Avoid-hardcoded-build-time-paths-in-the-output-binar.patch \
            file://0023-aarch64-configure-Pass-mcpu-along-with-march-to-dete.patch \
+           file://0024-qemu-stale-process.patch \
 "
 S = "${WORKDIR}/git"
 B = "${WORKDIR}/build-${TARGET_SYS}"
--

2.43.0

-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#197678): https://lists.openembedded.org/g/openembedded-core/message/197678
Mute This Topic: https://lists.openembedded.org/mt/105259855/1997914
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

#5 lore raw
Quoted 4 lines · Khem Raj

it would be better to disable them just for architectures where it fails since it seems to be a workaround, it would be good to keep it as small as possible

Initially, we implemented the changes to handle only for architectures where it fails and reviewed it with Richard in a meeting, but it was decided to skip the test cases rather than doing hard-coded changes.

Regards,
Yash

#6 lore raw
Quoted 11 lines · Yash via lists.openembedded.org

 On 02-04-2024 01:28, Khem Raj wrote:
 it would be better to disable them just for architectures where it

Quoted 4 lines

fails since it seems to be a workaround, it would be good to keep it
as small as possible

 Initially, we implemented the changes to handle only for architectures where it fails and reviewed it with Richard in a meeting,
 but it was decided to skip the test cases rather than doing hard- coded changes.

You could make the application of the patch conditional upon being the mips or powerpc architectures.

That does have advantages and disadvantages in that it is easy to miss when upgrading the glibc recipe though.

Cheers,

Richard

#7 lore raw
Quoted 9 lines · Richard Purdie
Quoted 6 lines · Yash via lists.openembedded.org

On 02-04-2024 01:28, Khem Raj wrote:
it would be better to disable them just for architectures where it

Quoted 4 lines

fails since it seems to be a workaround, it would be good to keep it
as small as possible

Quoted 8 lines
Quoted 5 lines

Initially, we implemented the changes to handle only for
architectures where it fails and reviewed it with Richard in a meeting,
but it was decided to skip the test cases rather than doing hard-
coded changes.

You could make the application of the patch conditional upon being the mips or powerpc architectures.

I think that will be more hassle than disabling it for all architectures.

Quoted 7 lines

That does have advantages and disadvantages in that it is easy to miss when upgrading the glibc recipe though.

Cheers,

Richard

#8 lore raw
Quoted 3 lines · Khem Raj
Quoted 2 lines

You could make the application of the patch conditional upon being the mips or powerpc architectures.

I think that will be more hassle than disabling it for all architectures.

I see that V2 of this patch is already merged with master branch. (https://git.yoctoproject.org/poky/commit/?id=8d27d8ff7cc65d2059ca29352b57301ffc35a163)
But, I have made the changes to apply the patch only for mips or ppc architectures.

Please let me know if you want me to send the patch. The changes are as follows:

meta/recipes-core/glibc/glibc_2.39.bb +0 -0
index 9122472689..497141f4bd 100644
@@ -49,7 +49,6 @@ SRC_URI = "${GLIBC_GIT_URI};branch=${SRCBRANCH};name=glibc \

file://0021-fix-create-thread-failed-in-unprivileged-process-BZ-.patch \
file://0022-Avoid-hardcoded-build-time-paths-in-the-output-binar.patch \
file://0023-aarch64-configure-Pass-mcpu-along-with-march-to-dete.patch \
- file://0024-qemu-stale-process.patch \
 "
 S = "${WORKDIR}/git"
 B = "${WORKDIR}/build-${TARGET_SYS}"
@@ -89,6 +88,11 @@ PACKAGECONFIG ??= "nscd memory-tagging" PACKAGECONFIG[nscd] = "--enable-nscd,--disable-nscd" PACKAGECONFIG[memory-tagging] = "--enable-memory-tagging,--disable-memory-tagging"

+python() {
+    if (d.getVar('MACHINE', True) == 'qemuppc' or d.getVar('MACHINE', True) == 'qemumips'):
+        d.appendVar('SRC_URI', ' file://0024-qemu-stale-process.patch')
+}
+
 do_patch:append() {
     bb.build.exec_func('do_fix_readlib_c', d)
 }

Regards,
Yash

Quoted 7 lines
Quoted 7 lines

That does have advantages and disadvantages in that it is easy to miss when upgrading the glibc recipe though.

Cheers,

Richard

#9 lore raw
Quoted 9 lines · Yash Shinde
Quoted 8 lines · Khem Raj
Quoted 2 lines

You could make the application of the patch conditional upon being the mips or powerpc architectures.

I think that will be more hassle than disabling it for all architectures.

Quoted 7 lines

 I see that V2 of this patch is already merged with master branch. (https://git.yoctoproject.org/poky/commit/?id=8d27d8ff7cc65d2059ca29352b57301ffc35a163)
 But, I have made the changes to apply the patch only for mips or ppc architectures.

 Please let me know if you want me to send the patch.

Your patch was disabling per machine, not per architecture which is different.

As Khem said, we're worried that disabling just for specific machines or architectures will make the recipe maintenance hard in the stable branch.

We're therefore happier with the original version which disables those two tests unconditionally which is why I merged that instead.

Cheers,

Richard