Skip to contentSkip to conversation

Support for Patchwork CI

PATCH Superseded Andrea Cervesato 3 participants 15 messages Apr 15, 2025 – Apr 16, 2025

v7 of 8 Superseded Latest is v8 →

lore
Share & tools
lore
#0 lore raw

v7 cover Support for Patchwork CI

Add support for patch-series validation in the patchwork ML. We use Github to schedule a trigger every 30 minutes, checking for new patche-series in parchwork which has not been tested yet.

The way we decide if a patch-series has been tested in patchwork, is by looking at its status (in particular, if it's "Needs Review / ACK"), as well as checking if test report has been uploaded to any of the series patches.

All communication to Patchwrok is done via REST API, using curl and js tools.

First, we create a script called patchwork-ci.sh that provides all the commands to read new untested patch-series, set their status and testing report. Then, we create a scheduled workflow in Gitlab, checking every 30 minutes if there are new untested patch-series. At the end, we trigger the main build workflow, used to validate LTP commits in our Github mainline. All the times we trigger the build workflow, we also provide the patch-series ID, that will be fetched and applied on the current branch before running the tests.

---
Changes in v7:
- patchwork.sh: fix bashism ==
- patchwork.sh: fix return code check for curl
- patchwork.sh: remove IFS= from loops
- patchwork.sh: local cmd in command_exists
- fix message for trigger scripts commit
- Link to v6: https://lore.kernel.org/r/[email protected]

Changes in v6:
- patchwork.sh: use more "[ $stuff ] || do_something" syntax
- patchwork.sh: remove set -ex
- patchwork.sh: variables default setup
- patchwork.sh: move file creation in tmp folder
- patchwork.sh: check multiple commands exist in once
- trigger workflow now triggers only on linux-test-project/ltp repo
- Link to v5: https://patchwork.ozlabs.org/project/ltp/list/?series=452320

Changes in v5:
- patchwork is now a utility that can be used separately
- Link to v4: https://lore.kernel.org/r/[email protected]

Changes in v4:
- make use of secrets to apply patches in order to set state to needs-review-ack
- check for curl failures and eventually exit 1
- use a different format for stored file in 'patchwork-ci.sh verify' command
- define PATCHWORK_CI_FILE
- some symbols like . are not supported by patchwork checks, so slurmify the string before sending checks context
- Link to v3: https://lore.kernel.org/r/[email protected]

Changes in v3:
- secrets are not supported by if: statement, so we remove it
- Link to v2: https://lore.kernel.org/r/[email protected]

Changes in v2:
- check for secrets.PATCHWORK_TOKEN before sending tests results
- v2 to re-trigger lore after missing 3/4 patch
- Link to v1: https://lore.kernel.org/r/[email protected]

---
Andrea Cervesato (4):
ci: install dependences for patchwork-ci script
ci: add patchwork communication script
ci: add ci-patchwork-trigger workflow
ci: apply patchwork series in ci-docker-build workflow

 .github/workflows/ci-docker-build.yml      |  39 ++++++-
 .github/workflows/ci-patchwork-trigger.yml |  65 +++++++++++
 ci/alpine-runtime.sh                       |   2 +
 ci/alpine.sh                               |   2 +
 ci/debian.i386.sh                          |   2 +
 ci/debian.sh                               |   2 +
 ci/fedora.sh                               |   2 +
 ci/tools/patchwork.sh                      | 173 +++++++++++++++++++++++++++++
 ci/tumbleweed.sh                           |   2 +
 9 files changed, 288 insertions(+), 1 deletion(-)

---
base-commit: 40c559787d15b2167ba657b3ba431e3299969d98
change-id: 20250410-patchwork_ci-7dc4ae02c40d

Best regards,

Signature · 5 lines
Andrea Cervesato <[email protected]>


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#1 lore raw

PATCH v7 1/4 ci: install dependences for patchwork-ci script

From: Andrea Cervesato <[email protected]>

---
 ci/alpine-runtime.sh | 2 ++
 ci/alpine.sh         | 2 ++
 ci/debian.i386.sh    | 2 ++
 ci/debian.sh         | 2 ++
 ci/fedora.sh         | 2 ++
 ci/tumbleweed.sh     | 2 ++
 6 files changed, 12 insertions(+)
ci/alpine-runtime.sh +2 -0
index 3bff42770bd8ae7da3e6e571a843e44c7a56181b..d0e1990d2c25c9f300dd47fb979af4d38e3ad3f9 100755
@@ -4,6 +4,8 @@
 
 apk add \
         acl \
+        curl \
+        jq \
         keyutils \
         libaio \
         libacl \
ci/alpine.sh +2 -0
index 5a44a6687b9eb7743c0fbbb288a133366b17da75..254f4aaece66ef177bc5b4a92cf5160443552378 100755
@@ -9,6 +9,8 @@ apk add \
 	autoconf \
 	automake \
 	clang \
+	curl \
+	jq \
 	gcc \
 	git \
 	acl-dev \
ci/debian.i386.sh +2 -0
index 2846053098033dd0215aa724738a90baff290430..44c7ddf2ff6f049ab847b374213c6aee74fbc130 100755
@@ -6,6 +6,8 @@ dpkg --add-architecture i386
 apt update
 
 apt install -y --no-install-recommends \
+	curl \
+	jq \
 	linux-libc-dev:i386 \
 	gcc-multilib \
 	libacl1-dev:i386 \
ci/debian.sh +2 -0
index f590b4b9a4f910c7e4229f4e2a818cf5b1e818bb..0445c92ec40819e49c0435b1881cf6b202d14087 100755
@@ -23,6 +23,8 @@ pkg_minimal="
 	debhelper
 	devscripts
 	clang
+	curl
+	jq
 	gcc
 	git
 	iproute2
ci/fedora.sh +2 -0
index bef5bcd2b519f4a32533a2d52f182c6dd5e1b7d0..494de928f7434f3310979e595e18162c2042f1e8 100755
@@ -9,6 +9,8 @@ $yum \
 	automake \
 	make \
 	clang \
+	curl \
+	jq \
 	gcc \
 	git \
 	findutils \
ci/tumbleweed.sh +2 -0
index 33937ec63a372ed92e8d4f05ecd435ce902fa5bb..d0607eed25f06fec3064ca75ec7875ce833f9d96 100755
@@ -8,6 +8,8 @@ $zyp \
 	autoconf \
 	automake \
 	clang \
+	curl \
+	jq \
 	findutils \
 	gcc \
 	git \
Signature · 5 lines
2.43.0


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#2 lore raw

PATCH v7 2/4 ci: add patchwork communication script

In reply to Andrea Cervesato

From: Andrea Cervesato <[email protected]>

Add a script to communicate with patchwork. Available commands are:

- state: change patch-series state
- check: send a tests report to patchwork
- verify: will print a list of new patch-series which has not been tested in the past hour (by default)

The script can be configured defining:

- PATCHWORK_URL: patchwork url to communicate with
- PATCHWORK_TOKEN: patchwork authentication token
- PATCHWORK_SINCE: timespan in seconds where we want to fetch patch-series

---
 ci/tools/patchwork.sh | 173 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 173 insertions(+)
ci/tools/patchwork.sh new file +173 -0
new file mode 100755
index 0000000000000000000000000000000000000000..176bd7737aacfc6178cd69a324d74f13bdc120e9
@@ -0,0 +1,173 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0-or-later
+#
+# Shell script to communicate with Patchwork via REST API.
+# It has been mainly created for CI purposes, but it can be used in the shell
+# by satisfying minimum requirements.
+#
+# Copyright (c) 2025 Andrea Cervesato <[email protected]>
+
+PATCHWORK_URL="${PATCHWORK_URL:-https://patchwork.ozlabs.org}"
+PATCHWORK_SINCE="${PATCHWORK_SINCE:-3600}"
+
+command_exists() {
+        local cmd=""
+
+        for cmd in "$@"; do
+                if ! command -v "$cmd" >/dev/null 2>&1; then
+                        echo "'$1' must be present in the system" >&2
+                        exit 1
+                fi
+        done
+}
+
+command_exists "curl" "jq"
+
+fetch_series() {
+        local current_time=$(date +%s)
+        local since_time=$(expr $current_time - $PATCHWORK_SINCE)
+        local date=$(date -u -d @$since_time +"%Y-%m-%dT%H:%M:%SZ")
+        local stdout=$(curl -k -G "$PATCHWORK_URL/api/events/" \
+                --data "category=series-completed" \
+                --data "project=ltp" \
+                --data "state=new" \
+                --data "since=$date" \
+                --data "archive=no")
+
+        [ $? -eq 0 ] || exit 1
+
+        echo "$stdout" | jq -r '.[] | "\(.payload.series.id) \(.payload.series.mbox)"'
+}
+
+get_patches() {
+        local series_id="$1"
+        local stdout="$(curl -k -G $PATCHWORK_URL/api/patches/ \
+                --data project=ltp \
+                --data series=$series_id)"
+
+        [ $? -eq 0 ] || exit 1
+
+        echo "$stdout" | jq -r '.[] | "\(.id)"'
+}
+
+verify_token_exists() {
+        if [ -z "$PATCHWORK_TOKEN" ]; then
+                echo "For this feature you need \$PATCHWORK_TOKEN"
+                exit 1
+        fi
+}
+
+set_patch_state() {
+        local patch_id="$1"
+        local state="$2"
+
+        verify_token_exists
+
+        curl -k -X PATCH \
+                -H "Authorization: Token $PATCHWORK_TOKEN" \
+                -F "state=$state" \
+                "$PATCHWORK_URL/api/patches/$patch_id/"
+
+        [ $? -eq 0 ] || exit 1
+}
+
+set_series_state() {
+        local series_id="$1"
+        local state="$2"
+
+        get_patches "$series_id" | while read -r patch_id; do
+                if [ "$patch_id" ]; then
+                        set_patch_state "$patch_id" "$state"
+                fi
+        done
+}
+
+get_checks() {
+        local patch_id="$1"
+        local stdout="$(curl -k -G $PATCHWORK_URL/api/patches/$patch_id/checks/)"
+
+        [ $? -eq 0 ] || exit 1
+
+        echo "$stdout" | jq -r '.[] | "\(.id)"'
+}
+
+already_tested() {
+        local series_id="$1"
+
+        get_patches "$series_id" | while read -r patch_id; do
+                [ "$patch_id" ] || continue
+
+                get_checks "$patch_id" | while read -r check_id; do
+                        if [ -n "$check_id" ]; then
+                                echo "$check_id"
+                                return
+                        fi
+                done
+        done
+}
+
+verify_new_patches() {
+        local tmp=$(mktemp -d)
+        local output="$tmp/series_ids.txt"
+
+        touch "$output"
+
+        fetch_series | while read -r series_id series_mbox; do
+                [ "$series_id" ] || continue
+
+                tested=$(already_tested "$series_id")
+                [ "$tested" ] && continue
+
+                echo "$series_id|$series_mbox" >>"$output"
+        done
+
+        cat "$output"
+}
+
+send_results() {
+        local series_id="$1"
+        local target_url="$2"
+
+        verify_token_exists
+
+        local context=$(echo "$3" | sed 's/:/_/g; s/\//-/g; s/\./-/g')
+
+        [ "$CC" ] && context="${context}_${CC}"
+        [ "$ARCH" ] && context="${context}_${ARCH}"
+
+        local result="$4"
+        [ "$result" = "cancelled" ] && return
+
+        local state="fail"
+        [ "$result" = "success" ] && state="success"
+
+        get_patches "$series_id" | while read -r patch_id; do
+                [ "$patch_id" ] || continue
+
+                curl -k -X POST \
+                        -H "Authorization: Token $PATCHWORK_TOKEN" \
+                        -F "state=$state" \
+                        -F "context=$context" \
+                        -F "target_url=$target_url" \
+                        -F "description=$result" \
+                        "$PATCHWORK_URL/api/patches/$patch_id/checks/"
+
+                [ $? -eq 0 ] && exit 1
+        done
+}
+
+case "$1" in
+state)
+        set_series_state "$2" "$3"
+        ;;
+check)
+        send_results "$2" "$3" "$4" "$5"
+        ;;
+verify)
+        verify_new_patches
+        ;;
+*)
+        echo "Available commands: state, check, verify" >&2
+        exit 1
+        ;;
+esac
Signature · 5 lines
2.43.0


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#12 lore raw

PATCH v7 3/4 ci: add ci-patchwork-trigger workflow

In reply to Andrea Cervesato

From: Andrea Cervesato <[email protected]>

Add ci-patchwork-trigger workflow that is meant to run every 15 minutes during the week and every 45 minutes during the weekend, checking for new untested LTP patches in the Mailing List and running the ci-docker-build workflow for them.

---
 .github/workflows/ci-patchwork-trigger.yml | 65 ++++++++++++++++++++++++++++++
 1 file changed, 65 insertions(+)
.github/workflows/ci-patchwork-trigger.yml new file +65 -0
new file mode 100644
index 0000000000000000000000000000000000000000..0938aed5ad0df8af8e2e575283b56475ff219ff3
@@ -0,0 +1,65 @@
+# Copyright (c) 2025 Andrea Cervesato <[email protected]>
+
+name: "Patchwork checker"
+on:
+  push:
+  schedule:
+    - cron: '*/15 * * * 1-5'
+    - cron: '*/45 * * * 6,0'
+
+env:
+  PATCHWORK_CI_FILE: patchwork-ci-output.txt
+
+jobs:
+  checker:
+    if: ${{ github.repository == 'linux-test-project/ltp' }}
+    runs-on: ubuntu-latest
+    steps:
+      - name: Git checkout
+        uses: actions/checkout@v1
+
+      - name: Verify new patches
+        id: verify
+        run: |
+          ./ci/tools/patchwork.sh verify > "$PATCHWORK_CI_FILE"
+          cat "$PATCHWORK_CI_FILE"
+
+      - name: Run tests
+        if: success()
+        uses: actions/github-script@v7
+        with:
+          script: |
+            const fs = require('fs');
+
+            const output = fs.readFileSync(process.env.PATCHWORK_CI_FILE, 'utf8');
+            if (output.length === 0) {
+              console.log("'patchwork-ci.sh verify' output is empty");
+              return;
+            }
+
+            const lines = output.split('\n');
+            if (lines.length === 0) {
+              console.log("No new patch-series found");
+              return;
+            }
+
+            for (const data of lines) {
+              const [series_id, series_mbox] = data.split('|');
+              if (series_id.length === 0 || series_mbox.length === 0) {
+                console.log(`Malformed data: ${data}`);
+                continue;
+              }
+
+              const response = await github.rest.actions.createWorkflowDispatch({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                ref: context.ref,
+                workflow_id: 'ci-docker-build.yml',
+                inputs: {
+                  SERIES_ID: series_id,
+                  SERIES_MBOX: series_mbox,
+                }
+              });
+
+              console.log(response);
+            }
Signature · 5 lines
2.43.0


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#13 lore raw

PATCH v7 4/4 ci: apply patchwork series in ci-docker-build workflow

In reply to Andrea Cervesato

From: Andrea Cervesato <[email protected]>

Modify ci-docker-build workflow in order to apply untested new patchwork patch-series inside the current branch and to send back results in the patchwork instance.

---
 .github/workflows/ci-docker-build.yml | 39 ++++++++++++++++++++++++++++++++++-
 1 file changed, 38 insertions(+), 1 deletion(-)
.github/workflows/ci-docker-build.yml +38 -1
index 44dcca055018a1fb06541a788e3152239aea2e66..b476d993c50a3369efdaa9a72787a7f006cfe0b9 100644
@@ -1,7 +1,19 @@
 # Copyright (c) 2021-2024 Petr Vorel <[email protected]>
 
 name: "Test building in various distros in Docker"
-on: [push, pull_request]
+on:
+  push:
+  pull_request:
+  workflow_dispatch:
+    inputs:
+      SERIES_ID:
+        description: LTP patch series ID
+        required: false
+        default: ''
+      SERIES_MBOX:
+        description: LTP patch series URL
+        required: false
+        default: ''
 
 permissions:
   contents: read # to fetch code (actions/checkout)
@@ -125,6 +137,20 @@ jobs:
     - name: Compiler version
       run: $CC --version
 
+    - name: Apply Patchwork series
+      if: inputs.SERIES_ID != '' && inputs.SERIES_MBOX != ''
+      env:
+        PATCHWORK_TOKEN: ${{ secrets.PATCHWORK_TOKEN }}
+      run: |
+        git config --global user.name 'GitHub CI'
+        git config --global user.email '[email protected]'
+        git config --global --add safe.directory "$GITHUB_WORKSPACE"
+
+        git checkout -b review_patch_series_"${{ inputs.SERIES_ID }}"
+        curl -k "${{ inputs.SERIES_MBOX }}" | git am
+
+        ./ci/tools/patchwork.sh state "${{ inputs.SERIES_ID }}" "needs-review-ack"
+
     - name: ver_linux
       run: ./ver_linux
 
@@ -158,3 +184,14 @@ jobs:
       run: |
         if [ "$MAKE_INSTALL" = 1 ]; then INSTALL_OPT="-i"; fi
         ./build.sh -r install -o ${TREE:-in} $INSTALL_OPT
+
+    - name: Send results to Patchwork
+      if: always() && inputs.SERIES_ID != '' && inputs.SERIES_MBOX != ''
+      env:
+        PATCHWORK_TOKEN: ${{ secrets.PATCHWORK_TOKEN }}
+      run: |
+        ./ci/tools/patchwork.sh check \
+          "${{ inputs.SERIES_ID }}" \
+          "${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}" \
+          "${{ matrix.container }}" \
+          "${{ job.status }}"
Signature · 5 lines
2.43.0


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#14 lore raw
In reply to Andrea Cervesato

Hi,

I temporarily disabled the schedule on my personal master branch, since this patch-set is going to be merged soon. FYI:

- we need a patchwork user to test new ML patches, so we can use its token
- we need to setup this token into the LTP settings, so we are able to communicate with patchwork

Kind regards,
Andrea Cervesato

Quoted 86 lines · Andrea Cervesato

Add support for patch-series validation in the patchwork ML. We use Github to schedule a trigger every 30 minutes, checking for new patche-series in parchwork which has not been tested yet.

The way we decide if a patch-series has been tested in patchwork, is by looking at its status (in particular, if it's "Needs Review / ACK"), as well as checking if test report has been uploaded to any of the series patches.

All communication to Patchwrok is done via REST API, using curl and js tools.

First, we create a script called patchwork-ci.sh that provides all the commands to read new untested patch-series, set their status and testing report. Then, we create a scheduled workflow in Gitlab, checking every 30 minutes if there are new untested patch-series. At the end, we trigger the main build workflow, used to validate LTP commits in our Github mainline. All the times we trigger the build workflow, we also provide the patch-series ID, that will be fetched and applied on the current branch before running the tests.

---
Changes in v7:
- patchwork.sh: fix bashism ==
- patchwork.sh: fix return code check for curl
- patchwork.sh: remove IFS= from loops
- patchwork.sh: local cmd in command_exists
- fix message for trigger scripts commit
- Link to v6: https://lore.kernel.org/r/[email protected]

Changes in v6:
- patchwork.sh: use more "[ $stuff ] || do_something" syntax
- patchwork.sh: remove set -ex
- patchwork.sh: variables default setup
- patchwork.sh: move file creation in tmp folder
- patchwork.sh: check multiple commands exist in once
- trigger workflow now triggers only on linux-test-project/ltp repo
- Link to v5: https://patchwork.ozlabs.org/project/ltp/list/?series=452320

Changes in v5:
- patchwork is now a utility that can be used separately
- Link to v4: https://lore.kernel.org/r/[email protected]

Changes in v4:
- make use of secrets to apply patches in order to set state to needs-review-ack
- check for curl failures and eventually exit 1
- use a different format for stored file in 'patchwork-ci.sh verify' command
- define PATCHWORK_CI_FILE
- some symbols like . are not supported by patchwork checks, so slurmify the string before sending checks context
- Link to v3: https://lore.kernel.org/r/[email protected]

Changes in v3:
- secrets are not supported by if: statement, so we remove it
- Link to v2: https://lore.kernel.org/r/[email protected]

Changes in v2:
- check for secrets.PATCHWORK_TOKEN before sending tests results
- v2 to re-trigger lore after missing 3/4 patch
- Link to v1: https://lore.kernel.org/r/[email protected]

---
Andrea Cervesato (4):
ci: install dependences for patchwork-ci script
ci: add patchwork communication script
ci: add ci-patchwork-trigger workflow
ci: apply patchwork series in ci-docker-build workflow

  .github/workflows/ci-docker-build.yml      |  39 ++++++-
  .github/workflows/ci-patchwork-trigger.yml |  65 +++++++++++
  ci/alpine-runtime.sh                       |   2 +
  ci/alpine.sh                               |   2 +
  ci/debian.i386.sh                          |   2 +
  ci/debian.sh                               |   2 +
  ci/fedora.sh                               |   2 +
  ci/tools/patchwork.sh                      | 173 +++++++++++++++++++++++++++++
  ci/tumbleweed.sh                           |   2 +
  9 files changed, 288 insertions(+), 1 deletion(-)

---
base-commit: 40c559787d15b2167ba657b3ba431e3299969d98
change-id: 20250410-patchwork_ci-7dc4ae02c40d

Best regards,

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#3 lore raw

Re: [LTP] [PATCH v7 2/4] ci: add patchwork communication script

In reply to Andrea Cervesato

Hi Andrea,

...

ci/tools/patchwork.sh line 37

Quoted 12 lines
ci/tools/patchwork.sh new file +12 -0
+fetch_series() {
+        local current_time=$(date +%s)
+        local since_time=$(expr $current_time - $PATCHWORK_SINCE)
+        local date=$(date -u -d @$since_time +"%Y-%m-%dT%H:%M:%SZ")
+        local stdout=$(curl -k -G "$PATCHWORK_URL/api/events/" \
+                --data "category=series-completed" \
+                --data "project=ltp" \
+                --data "state=new" \
+                --data "since=$date" \
+                --data "archive=no")
+
+        [ $? -eq 0 ] || exit 1

You may have noticed in tst_test.sh, that local variable never uses $(...). It assign single value, but never call $(...). This is for a reason.

Try:

$ cat foo.sh
#!/bin/sh

foo()

{
    local foo=$(aasfd_command_which_fails)

    [ $? -eq 0 ] || exit 1
    echo "run after"
}

bar()

{
    local foo

    foo=$(aasfd_command_which_fails)

    [ $? -eq 0 ] || exit 1
    echo "will not run after"
}

foo
bar

---

$ bash ./foo.sh
./foo.sh: line 5: aasfd_command_which_fails: command not found
run after
./foo.sh: line 15: aasfd_command_which_fails: command not found

What happen? $? is assigned from result of local keyword, it overwrite previous result from $(...). Note even '#!/bin/sh -e' would not cause it to fail early.

(Deliberately test with bash to demonstrate local behaves oddly not even in dash or 'busybox sh' but even with bash. And yes, given how many errors we caught with this script and generate_arch.sh and generate_syscalls.sh due shell strange syntax and behavior makes me wonder if we really want to use shell scripts for anything longer than 5 lines.)

Kind regards,
Petr

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#4 lore raw

Re: [LTP] [PATCH v7 2/4] ci: add patchwork communication script

Quoted 7 lines · Petr Vorel

Hi Andrea,
You may have noticed in tst_test.sh, that local variable never uses $(...). It assign single value, but never call $(...). This is for a reason. [...]
What happen? $? is assigned from result of local keyword,
it overwrite previous result from $(...). Note even '#!/bin/sh -e' would not cause it to fail early.

Here is the corresponding documentation:
https://www.gnu.org/savannah-checkouts/gnu/bash/manual/bash.html#index-local

Quoted 1 line

The return status is zero unless local is used outside a function, an

invalid name is supplied, or name is a readonly variable

So:
If you are interested in return statuses then don't *mask them* with local.

Use this pattern instead:

local my_var
my_var=$(my_expr)

Kind regards,
Dennis

Quoted 9 lines

(Deliberately test with bash to demonstrate local behaves oddly not even in dash or 'busybox sh' but even with bash. And yes, given how many errors we caught with this script and generate_arch.sh and generate_syscalls.sh due shell strange syntax and behavior makes me wonder if we really want to use shell scripts for anything longer than 5 lines.)

Kind regards,
Petr

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#8 lore raw

Re: [LTP] [PATCH v7 2/4] ci: add patchwork communication script

In reply to Andrea Cervesato
Quoted 16 lines · Andrea Cervesato

From: Andrea Cervesato <[email protected]>

Add a script to communicate with patchwork. Available commands are:

- state: change patch-series state
- check: send a tests report to patchwork
- verify: will print a list of new patch-series which has not been tested in the past hour (by default)

The script can be configured defining:

- PATCHWORK_URL: patchwork url to communicate with
- PATCHWORK_TOKEN: patchwork authentication token
- PATCHWORK_SINCE: timespan in seconds where we want to fetch patch-series

Quoted 7 lines
---
 ci/tools/patchwork.sh | 173

++++++++++++++++++++++++++++++++++++++++++++++++++

Maybe better to move this independent tool to the script/ dir? This ci/tools/ is so hidden that it can easily be overlooked.

Signature · 5 lines
Regards,
Li Wang

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#6 lore raw

Re: [LTP] [PATCH v7 2/4] ci: add patchwork communication script

In reply to Petr Vorel

Hi Petr,

this is exactly the reason why I prefer "set -e" (I'm not a shell script expert anyway...). But I was actually thinking that the whole pipe command will fail with !=0 ret code anyway, since any curl output which is not a REST API reply (JSON string in particular) and having the right format, will cause jq to fail. So the first implementation is technically working. I would like to keep that simple and check $? later on after the piped command as it was before. Let me know what you think.

- Andrea

Quoted 62 lines · Petr Vorel

Hi Andrea,

...

ci/tools/patchwork.sh line 37

Quoted 12 lines
ci/tools/patchwork.sh new file +12 -0
+fetch_series() {
+        local current_time=$(date +%s)
+        local since_time=$(expr $current_time - $PATCHWORK_SINCE)
+        local date=$(date -u -d @$since_time +"%Y-%m-%dT%H:%M:%SZ")
+        local stdout=$(curl -k -G "$PATCHWORK_URL/api/events/" \
+                --data "category=series-completed" \
+                --data "project=ltp" \
+                --data "state=new" \
+                --data "since=$date" \
+                --data "archive=no")
+
+        [ $? -eq 0 ] || exit 1

You may have noticed in tst_test.sh, that local variable never uses $(...). It assign single value, but never call $(...). This is for a reason.

Try:

$ cat foo.sh
#!/bin/sh

foo()

{
     local foo=$(aasfd_command_which_fails)

     [ $? -eq 0 ] || exit 1
     echo "run after"
}

bar()

{
     local foo

     foo=$(aasfd_command_which_fails)

     [ $? -eq 0 ] || exit 1
     echo "will not run after"
}

foo
bar

---

$ bash ./foo.sh
./foo.sh: line 5: aasfd_command_which_fails: command not found
run after
./foo.sh: line 15: aasfd_command_which_fails: command not found

What happen? $? is assigned from result of local keyword, it overwrite previous result from $(...). Note even '#!/bin/sh -e' would not cause it to fail early.

(Deliberately test with bash to demonstrate local behaves oddly not even in dash or 'busybox sh' but even with bash. And yes, given how many errors we caught with this script and generate_arch.sh and generate_syscalls.sh due shell strange syntax and behavior makes me wonder if we really want to use shell scripts for anything longer than 5 lines.)

Kind regards,
Petr

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#7 lore raw

Re: [LTP] [PATCH v7 2/4] ci: add patchwork communication script

In reply to Petr Vorel

Hi Petr,

Quoted 8 lines · Petr Vorel

(Deliberately test with bash to demonstrate local behaves oddly not even in dash or 'busybox sh' but even with bash. And yes, given how many errors we caught with this script and generate_arch.sh and generate_syscalls.sh due shell strange syntax and behavior makes me wonder if we really want to use shell scripts for anything longer than 5 lines.)

Kind regards,
Petr

Not a big deal, the script is really simple anyway. But I will put on TODO list the conversion of this script to python 3.x. Thanks for helping with shell scripting.

- Andrea

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#9 lore raw

Re: [LTP] [PATCH v7 2/4] ci: add patchwork communication script

In reply to Li Wang via ltp

Hi Li,

Quoted 2 lines · Li Wang

Maybe better to move this independent tool to the script/ dir? This ci/tools/ is so hidden that it can easily be overlooked.

We thought about it, but since it's mostly a related CI script we decided to move it into ci/tools. The scripts/ folder seems to be more related to LTP tests development, so I'm a bit puzzled now :-)

I let Petr Vorel the last word about it

- Andrea

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#10 lore raw

Re: [LTP] [PATCH v7 2/4] ci: add patchwork communication script

Hi Li, Andrea,

Quoted 11 lines

Hi Li,

Quoted 2 lines · Li Wang

Maybe better to move this independent tool to the script/ dir? This ci/tools/ is so hidden that it can easily be overlooked.

We thought about it, but since it's mostly a related CI script we decided to move it into ci/tools. The scripts/ folder seems to be more related to LTP tests development, so I'm a bit puzzled now :-)

I let Petr Vorel the last word about it

Li, FYI I asked Andrea to keep patchwork.sh script in ci/tools/ because it's related to CI and probably nobody will use it directly. I would like to keep in script/ only the scripts which will be used by people. If you really think that somebody will be using it, we can have it in script/. WDYT?

(And why in ci/tools/ and not in ci/? I'd like to keep ci/*.sh for LTP users as an easy way to install the dependencies.)

Kind regards,
Petr

Quoted 1 line

- Andrea

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#11 lore raw

Re: [LTP] [PATCH v7 2/4] ci: add patchwork communication script

Quoted 25 lines · Petr Vorel

Hi Li, Andrea,

Quoted 7 lines

Hi Li,

Quoted 2 lines · Li Wang

Maybe better to move this independent tool to the script/ dir? This ci/tools/ is so hidden that it can easily be overlooked.

We thought about it, but since it's mostly a related CI script we

decided to

Quoted 1 line

move it into ci/tools. The scripts/ folder seems to be more related to

LTP

Quoted 3 lines

tests development, so I'm a bit puzzled now :-)

I let Petr Vorel the last word about it

Li, FYI I asked Andrea to keep patchwork.sh script in ci/tools/ because it's
related to CI and probably nobody will use it directly. I would like to keep in
script/ only the scripts which will be used by people. If you really think that
somebody will be using it, we can have it in script/. WDYT?

Quoted 4 lines

(And why in ci/tools/ and not in ci/? I'd like to keep ci/*.sh for LTP users
as an easy way to install the dependencies.)

Sure, I'm OK with putting it in the ci/tools/.

Signature · 5 lines
Regards,
Li Wang

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#5 lore raw

Re: [LTP] [PATCH v7 2/4] ci: add patchwork communication script

In reply to Dennis Brendel via ltp

Hi Dennis,

Quoted 15 lines · Petr Vorel
Quoted 7 lines

Hi Andrea,
You may have noticed in tst_test.sh, that local variable never uses $(...). It assign single value, but never call $(...). This is for a reason. [...]
What happen? $? is assigned from result of local keyword,
it overwrite previous result from $(...). Note even '#!/bin/sh -e' would not cause it to fail early.

Here is the corresponding documentation:
https://www.gnu.org/savannah-checkouts/gnu/bash/manual/bash.html#index-local

Quoted 1 line

The return status is zero unless local is used outside a function, an

invalid name is supplied, or name is a readonly variable

Quoted 7 lines

So:
If you are interested in return statuses then don't *mask them* with local.

Use this pattern instead:

local my_var
my_var=$(my_expr)

Thanks for much better explanation than I provided + link to the doc. I'll Cc you on my shell related patches :).

I remember this since 87a82a62ce ("lib/tst_test.sh: fix ROD_SILENT command return status check") [1], another related is e267a022cd ("tst_test.sh: Fix $@ usage on older dash releases") [2].

Because shell has quirks (both POSIX portable syntax or bash or whatever shell extension) we've had many fixes for shell code, e.g. 0c0076fbaf ("tst_test.sh:
Fix ROD_SILENT() quoting") [3], although some of them are non-bash specific, e.g. 0bb01e67b3 ("shell: Fix handling default parameters in tst_mkfs()") [4] (just to list very few).

The biggest problem is that some things aren't easily done with plain shell, that's why we need to use C helpers in testcases/lib (some of them are just reusing C API to avoid reimplementing code, but e.g. testcases/lib/tst_timeout_kill.c was really needed after few attempts to write code in bash). Also cgroup tests aren't reliable when written in shell. Therefore we're trying to get rid of shell API (rewritten tests into C, or make them to use testcases/lib/tst_run_shell.c, which makes the layer really thin).

Kind regards,
Petr

[1] https://github.com/linux-test-project/ltp/commit/87a82a62ce3fcdaf174d6e4529e0324742e13684
[2] https://github.com/linux-test-project/ltp/commit/e267a022cd2deb0d5e7280570c711420927ad817
[3] https://github.com/linux-test-project/ltp/commit/0c0076fbaf6e0059b470fadff6240fc56952c218
[4] https://github.com/linux-test-project/ltp/commit/0bb01e67b3ba079f8f069a99422d783fe6da4287

Quoted 11 lines

Kind regards,
Dennis

Quoted 8 lines

(Deliberately test with bash to demonstrate local behaves oddly not even in dash or 'busybox sh' but even with bash. And yes, given how many errors we caught with this script and generate_arch.sh and generate_syscalls.sh due shell strange syntax and behavior makes me wonder if we really want to use shell scripts for anything longer than 5 lines.)

Kind regards,
Petr

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp