git 2.24: git revert <commit1> <commit2> requires extra '--continue'?
Hi! I'm using git 2.24 (or, a variant of that packaged my distro -- I can try to build my own if this is deemed not reproducible), and I feel like I've been seeing a regression here:
Previously, when reverting multiple commits (with the default --edit behavior), as soon as I'm done editing the first revert commit message, the second revert commit pops up an editor, and I'm on my way. Now, it seems to require an extra 'git revert --continue' prompt in between, yet it doesn't actually recommend that. This is highly confusing, and seemingly unnecessary.
An annotated transcript provided below.
Note that none of this happens with --no-edit; my reverts happen smoothly, with no extra need for --continue.
Regards,
Brian
$ mkdir tmp $ cd tmp /tmp$ git init Initialized empty Git repository in [...]/tmp/.git/ /tmp$ touch foo /tmp$ git add foo /tmp$ echo bar >> foo /tmp$ git commit -am baz [master (root-commit) a388f78d9013] baz 1 file changed, 1 insertion(+) create mode 100644 foo /tmp$ echo pow >> foo /tmp$ git commit -am whizzbang [master 51753222dd9a] whizzbang 1 file changed, 1 insertion(+) /tmp$ echo pop >> foo /tmp$ git commit -am nothing [master 7153607b11e0] nothing 1 file changed, 1 insertion(+) /tmp$ git revert HEAD HEAD^ ## EDITOR pops up, as expected [master 586469974ec2] Revert "nothing" 1 file changed, 1 deletion(-) On branch master Revert currently in progress.
nothing to commit, working tree clean
## Unexpected, confusing pause? No prompt to '--continue' /tmp$ git log --oneline586469974ec2 (HEAD -> master) Revert "nothing" 7153607b11e0 nothing51753222dd9a whizzbanga388f78d9013 baz
/tmp$ git status
On branch master
Revert currently in progress.
(run "git revert --continue" to continue)
(use "git revert --skip" to skip this patch) (use "git revert --abort" to cancel the revert operation)
nothing to commit, working tree clean
/tmp$ git revert --continue
## EDITOR pops up, as expected
[master b8dd23f61d07] Revert "whizzbang"
1 file changed, 1 deletion(-)
On branch master
Revert currently in progress.
nothing to commit, working tree clean
## Unexpected state; both reverts happened, but revert is still in progress? /tmp$ git log --onelineb8dd23f61d07 (HEAD -> master) Revert "whizzbang"586469974ec2 Revert "nothing"7153607b11e0 nothing51753222dd9a whizzbanga388f78d9013 baz
/tmp$ git status
On branch master
Revert currently in progress.
(run "git revert --continue" to continue)
(use "git revert --skip" to skip this patch)
(use "git revert --abort" to cancel the revert operation)
nothing to commit, working tree clean
## OK...I'll run it one more time.
/tmp$ git revert --continue
/tmp$ git status
On branch master
nothing to commit, working tree clean
## *Now* I'm done
/tmp$ git log --onelineb8dd23f61d07 (HEAD -> master) Revert "whizzbang"586469974ec2 Revert "nothing"7153607b11e0 nothing51753222dd9a whizzbanga388f78d9013 baz
Quoted 10 lines · Brian Norris
Hi! I'm using git 2.24 (or, a variant of that packaged my distro -- I can try to build my own if this is deemed not reproducible), and I feel like I've been seeing a regression here:
Previously, when reverting multiple commits (with the default --edit behavior), as soon as I'm done editing the first revert commit message, the second revert commit pops up an editor, and I'm on my way. Now, it seems to require an extra 'git revert --continue' prompt in between, yet it doesn't actually recommend that. This is highly confusing, and seemingly unnecessary.
Thanks for the report, this is indeed a regression in v2.24.0: it bisects down to a47ba3c777 (rebase -i: check for updated todo after squash and reword, 2019-08-19) [Cc'ing Phillip]. It's not specific to 'git revert', but with a slight variation affects 'git cherry-pick' as well.
Quoted 1 line
An annotated transcript provided below.
I transcribed your transcript into tests that can be run in our test framework and demonstrate this regression:
--- >8 ---
#!/bin/sh
test_description='test'
. ./test-lib.sh
test_expect_success "Brian's revert regression" '
test_create_repo revert && ( cd revert && echo 1 >file && git add file && git commit -m first && echo 2 >file && git commit -am second && echo 3 >file && git commit -am third && git checkout -b branch && git revert --edit HEAD HEAD^ && echo 1 >expect && test_cmp expect file )
'
test_expect_success "a variant of Brian's regression for cherry-pick" '
test_create_repo cherry-pick && ( cd cherry-pick && echo 1 >file && git add file && git commit -m first && echo 2 >file && git commit -am second && echo 3 >file && git commit -am third && git checkout -b branch HEAD^^ && git cherry-pick --edit master^ master && echo 3 >expect && test_cmp expect file )
'
test_done
--- >8 ---
They both succeed on a47ba3c777's parent, but fail on a47ba3c777 when the 'git revert' or 'git cherry-pick' commands return with exit code 1 after reverting/cherry-picking the first commit instead of processing the second commit:
+ git revert --edit HEAD HEAD^
[branch 88ea48c] Revert "third"
Author: A U Thor <[email protected]>
1 file changed, 1 insertion(+), 1 deletion(-)
On branch branch
Revert currently in progress.
nothing to commit, working tree clean
error: last command exited with $?=1
not ok 1 - Brian's revert regression
+ git cherry-pick --edit master^ master
[branch 2cb3f74] second
Author: A U Thor <[email protected]>
Date: Sat Nov 23 00:17:32 2019 +0000
1 file changed, 1 insertion(+), 1 deletion(-)
On branch branch
Cherry-pick currently in progress.
nothing to commit, working tree clean
The previous cherry-pick is now empty, possibly due to conflict resolution.
If you wish to commit it anyway, use:
git commit --allow-empty
If you wish to skip this commit, use:
git reset
Then "git cherry-pick --continue" will resume cherry-picking the remaining commits.
error: last command exited with $?=1
not ok 2 - a variant of Brian's regression for cherry-pick
These test should probably be squeezed into 't3508-cherry-pick-many-commits.sh', but Friday has just turned into Saturday around here, so that's enough from me for now.
Quoted 82 lines
Note that none of this happens with --no-edit; my reverts happen smoothly, with no extra need for --continue.
Regards,
Brian$ mkdir tmp $ cd tmp /tmp$ git init Initialized empty Git repository in [...]/tmp/.git/ /tmp$ touch foo /tmp$ git add foo /tmp$ echo bar >> foo /tmp$ git commit -am baz [master (root-commit) a388f78d9013] baz 1 file changed, 1 insertion(+) create mode 100644 foo /tmp$ echo pow >> foo /tmp$ git commit -am whizzbang [master 51753222dd9a] whizzbang 1 file changed, 1 insertion(+) /tmp$ echo pop >> foo /tmp$ git commit -am nothing [master 7153607b11e0] nothing 1 file changed, 1 insertion(+) /tmp$ git revert HEAD HEAD^ ## EDITOR pops up, as expected [master 586469974ec2] Revert "nothing" 1 file changed, 1 deletion(-) On branch master Revert currently in progress.nothing to commit, working tree clean
## Unexpected, confusing pause? No prompt to '--continue' /tmp$ git log --oneline586469974ec2(HEAD -> master) Revert "nothing"7153607b11e0nothing51753222dd9awhizzbanga388f78d9013baz
/tmp$ git status
On branch master
Revert currently in progress.
(run "git revert --continue" to continue)
(use "git revert --skip" to skip this patch) (use "git revert --abort" to cancel the revert operation)nothing to commit, working tree clean
/tmp$ git revert --continue
## EDITOR pops up, as expected
[masterb8dd23f61d07] Revert "whizzbang"1 file changed, 1 deletion(-)On branch master
Revert currently in progress.nothing to commit, working tree clean
## Unexpected state; both reverts happened, but revert is still in progress? /tmp$ git log --onelineb8dd23f61d07(HEAD -> master) Revert "whizzbang"586469974ec2Revert "nothing"7153607b11e0nothing51753222dd9awhizzbanga388f78d9013baz
/tmp$ git status
On branch master
Revert currently in progress.
(run "git revert --continue" to continue)
(use "git revert --skip" to skip this patch)
(use "git revert --abort" to cancel the revert operation)nothing to commit, working tree clean
## OK...I'll run it one more time.
/tmp$ git revert --continue
/tmp$ git status
On branch master
nothing to commit, working tree clean
## *Now* I'm done
/tmp$ git log --onelineb8dd23f61d07(HEAD -> master) Revert "whizzbang"586469974ec2Revert "nothing"7153607b11e0nothing51753222dd9awhizzbanga388f78d9013baz
Hi Brian & Gábor
Thanks to Brian for reporting this and Gábor for bisecting and creating the tests.
Quoted 12 lines · SZEDER Gábor
Quoted 10 lines · Brian Norris
Hi! I'm using git 2.24 (or, a variant of that packaged my distro -- I can try to build my own if this is deemed not reproducible), and I feel like I've been seeing a regression here:
Previously, when reverting multiple commits (with the default --edit behavior), as soon as I'm done editing the first revert commit message, the second revert commit pops up an editor, and I'm on my way. Now, it seems to require an extra 'git revert --continue' prompt in between, yet it doesn't actually recommend that. This is highly confusing, and seemingly unnecessary.
Quoted 5 lines
Thanks for the report, this is indeed a regression in v2.24.0: it bisects down to
a47ba3c777(rebase -i: check for updated todo after squash and reword, 2019-08-19) [Cc'ing Phillip]. It's not specific to 'git revert', but with a slight variation affects 'git cherry-pick' as well.
Thanks, I suspect it's missing an 'if (is_rebase_i(opts))' I'll take a look later and come up with a fix
Best Wishes
Phillip
Quoted 191 lines
Quoted 1 line
An annotated transcript provided below.
I transcribed your transcript into tests that can be run in our test framework and demonstrate this regression:
--- >8 ---#!/bin/sh
test_description='test'
. ./test-lib.sh
test_expect_success"Brian's revert regression" 'test_create_repo revert && ( cd revert && echo 1 >file && git add file && git commit -m first && echo 2 >file && git commit -am second && echo 3 >file && git commit -am third && git checkout -b branch && git revert --edit HEAD HEAD^ && echo 1 >expect && test_cmp expect file )'
test_expect_success"a variant of Brian's regression for cherry-pick" 'test_create_repo cherry-pick && ( cd cherry-pick && echo 1 >file && git add file && git commit -m first && echo 2 >file && git commit -am second && echo 3 >file && git commit -am third && git checkout -b branch HEAD^^ && git cherry-pick --edit master^ master && echo 3 >expect && test_cmp expect file )'
test_done--- >8 ---They both succeed on a47ba3c777's parent, but fail on
a47ba3c777when the 'git revert' or 'git cherry-pick' commands return with exit code 1 after reverting/cherry-picking the first commit instead of processing the second commit:+ git revert --edit HEAD HEAD^
[branch 88ea48c] Revert "third"
Author: A U Thor <[email protected]>
1 file changed, 1 insertion(+), 1 deletion(-)
On branch branch
Revert currently in progress.nothing to commit, working tree clean
error: last command exited with $?=1
not ok 1 - Brian's revert regression+ git cherry-pick --edit master^ master
[branch 2cb3f74] second
Author: A U Thor <[email protected]>
Date: Sat Nov 23 00:17:32 2019 +0000
1 file changed, 1 insertion(+), 1 deletion(-)
On branch branch
Cherry-pick currently in progress.nothing to commit, working tree clean
The previous cherry-pick is now empty, possibly due to conflict resolution.
If you wish to commit it anyway, use:git commit --allow-emptyIf you wish to skip this commit, use:
git reset
Then "git cherry-pick --continue" will resume cherry-picking the remaining commits.
error: last command exited with $?=1
not ok 2 - a variant of Brian's regression for cherry-pickThese test should probably be squeezed into '
t3508-cherry-pick-many-commits.sh', but Friday has just turned into Saturday around here, so that's enough from me for now.Quoted 82 lines
Note that none of this happens with --no-edit; my reverts happen smoothly, with no extra need for --continue.
Regards,
Brian$ mkdir tmp $ cd tmp /tmp$ git init Initialized empty Git repository in [...]/tmp/.git/ /tmp$ touch foo /tmp$ git add foo /tmp$ echo bar >> foo /tmp$ git commit -am baz [master (root-commit) a388f78d9013] baz 1 file changed, 1 insertion(+) create mode 100644 foo /tmp$ echo pow >> foo /tmp$ git commit -am whizzbang [master 51753222dd9a] whizzbang 1 file changed, 1 insertion(+) /tmp$ echo pop >> foo /tmp$ git commit -am nothing [master 7153607b11e0] nothing 1 file changed, 1 insertion(+) /tmp$ git revert HEAD HEAD^ ## EDITOR pops up, as expected [master 586469974ec2] Revert "nothing" 1 file changed, 1 deletion(-) On branch master Revert currently in progress.nothing to commit, working tree clean
## Unexpected, confusing pause? No prompt to '--continue' /tmp$ git log --oneline586469974ec2(HEAD -> master) Revert "nothing"7153607b11e0nothing51753222dd9awhizzbanga388f78d9013baz
/tmp$ git status
On branch master
Revert currently in progress.
(run "git revert --continue" to continue)
(use "git revert --skip" to skip this patch) (use "git revert --abort" to cancel the revert operation)nothing to commit, working tree clean
/tmp$ git revert --continue
## EDITOR pops up, as expected
[masterb8dd23f61d07] Revert "whizzbang"1 file changed, 1 deletion(-)On branch master
Revert currently in progress.nothing to commit, working tree clean
## Unexpected state; both reverts happened, but revert is still in progress? /tmp$ git log --onelineb8dd23f61d07(HEAD -> master) Revert "whizzbang"586469974ec2Revert "nothing"7153607b11e0nothing51753222dd9awhizzbanga388f78d9013baz
/tmp$ git status
On branch master
Revert currently in progress.
(run "git revert --continue" to continue)
(use "git revert --skip" to skip this patch)
(use "git revert --abort" to cancel the revert operation)nothing to commit, working tree clean
## OK...I'll run it one more time.
/tmp$ git revert --continue
/tmp$ git status
On branch master
nothing to commit, working tree clean
## *Now* I'm done
/tmp$ git log --onelineb8dd23f61d07(HEAD -> master) Revert "whizzbang"586469974ec2Revert "nothing"7153607b11e0nothing51753222dd9awhizzbanga388f78d9013baz
PATCH 1/1 sequencer: don't re-read todo for revert and cherry-pick
When 'git revert' or 'git cherry-pick --edit' is invoked with multiple commits, then after editing the first commit message is finished both these commands should continue with processing the second commit and launch another editor for its commit message, assuming there are no conflicts, of course.
Alas, this inadvertently changed with commit a47ba3c777 (rebase -i:
check for updated todo after squash and reword, 2019-08-19): after editing the first commit message is finished, both 'git revert' and 'git cherry-pick --edit' exit with error, claiming that "nothing to commit, working tree clean".
The reason for the changed behaviour is twofold:
- Prior to a47ba3c777 the up-to-dateness of the todo list file was only checked after 'exec' instructions, and that commit moved those checks to the common code path. The intention was that this check should be performed after instructions spawning an editor ('squash' and 'reword') as well, so the ongoing 'rebase -i' notices when the user runs a 'git rebase --edit-todo' while squashing/rewording a commit message.
However, as it happened that check is now performed even after 'revert' and 'pick' instructions when they involved editing the commit message. And 'revert' by default while 'pick' optionally (with 'git cherry-pick --edit') involves editing the commit message.
- When invoking 'git revert' or 'git cherry-pick --edit' with multiple commits they don't read a todo list file but assemble the todo list in memory, thus the associated stat data used to check whether the file has been updated is all zeroed out initially.
Then the sequencer writes all instructions (including the very first) to the todo file, executes the first 'revert/pick' instruction, and after the user finished editing the commit message the changes of a47ba3c777 kick in, and it checks whether the todo file has been modified. The initial all-zero stat data obviously differs from the todo file's current stat data, so the sequencer concludes that the file has been modified. Technically it is not wrong, of course, because the file just has been written indeed by the sequencer itself, though the file's contents still match what the sequencer was invoked with in the beginning. Consequently, after re-reading the todo file the sequencer executes the same first instruction _again_, thus ending up in that "nothing to commit" situation.
The todo list was never meant to be edited during multi-commit 'git revert' or 'cherry-pick' operations, so perform that "has the todo file been modified" check only when the sequencer was invoked as part of an interactive rebase.
---
Quoted 2 lines · Phillip Wood
Thanks, I suspect it's missing an 'if (
is_rebase_i(opts))' I'll take a look later and come up with a fix
That missing condition was my hunch yesterday evening as well, and while it did fix my tests and didn't break anything else, it took some time to wrap my head around some of the subtleties that are going on in the sequencer. That's why the commit message ended up this long as it did.
In the end I decided to add the new tests to 't3429-rebase-edit-todo.sh', because, though these new tests don't actually check 'rebase', that is the one test script focusing on (re-)reading the todo file in particular.
sequencer.c | 2 +- t/t3429-rebase-edit-todo.sh | 30 ++++++++++++++++++++++++++++++ 2 files changed, 31 insertions(+), 1 deletion(-)
sequencer.c +1 -1
@@ -3791,7 +3791,7 @@ static int pick_commits(struct repository *r, item->commit, arg, item->arg_len, opts, res, 0); - }else if (check_todo && !res) { + }else if (is_rebase_i (opts) && check_todo && !res) {struct stat st;if (stat (get_todo_path (opts), &st)) {
t/t3429-rebase-edit-todo.sh +30 -0
@@ -52,4 +52,34 @@ test_expect_success 'todo is re-read after reword and squash' ' test_cmp expected actual ' +test_expect_success 're-reading todo doesnt interfere with revert --edit' ' + git reset --hard third && + + git revert --edit third second && + + cat >expect <<-\EOF && + Revert "second" + Revert "third" + third + second + first + EOF + git log --format="%s" >actual && + test_cmp expect actual +' + +test_expect_success 're-reading todo doesnt interfere with cherry-pick --edit' ' + git reset --hard first && + + git cherry-pick --edit second third && + + cat >expect <<-\EOF && + third + second + first + EOF + git log --format="%s" >actual && + test_cmp expect actual +' + test_done
Signature · 1 line
2.24.0.532.ge18579ded8
Re: [PATCH] sequencer: don't re-read todo for revert and cherry-pick
Hi,
Quoted 47 lines · SZEDER Gábor
When 'git revert' or 'git cherry-pick --edit' is invoked with multiple commits, then after editing the first commit message is finished both these commands should continue with processing the second commit and launch another editor for its commit message, assuming there are no conflicts, of course.
Alas, this inadvertently changed with commit
a47ba3c777(rebase -i:
check for updated todo after squash and reword, 2019-08-19): after editing the first commit message is finished, both 'git revert' and 'git cherry-pick --edit' exit with error, claiming that "nothing to commit, working tree clean".The reason for the changed behaviour is twofold:
- Prior to
a47ba3c777the up-to-dateness of the todo list file was only checked after 'exec' instructions, and that commit moved those checks to the common code path. The intention was that this check should be performed after instructions spawning an editor ('squash' and 'reword') as well, so the ongoing 'rebase -i' notices when the user runs a 'git rebase --edit-todo' while squashing/rewording a commit message.However, as it happened that check is now performed even after 'revert' and 'pick' instructions when they involved editing the commit message. And 'revert' by default while 'pick' optionally (with 'git cherry-pick --edit') involves editing the commit message.
- When invoking 'git revert' or 'git cherry-pick --edit' with multiple commits they don't read a todo list file but assemble the todo list in memory, thus the associated stat data used to check whether the file has been updated is all zeroed out initially.
Then the sequencer writes all instructions (including the very first) to the todo file, executes the first 'revert/pick' instruction, and after the user finished editing the commit message the changes of
a47ba3c777kick in, and it checks whether the todo file has been modified. The initial all-zero stat data obviously differs from the todo file's current stat data, so the sequencer concludes that the file has been modified. Technically it is not wrong, of course, because the file just has been written indeed by the sequencer itself, though the file's contents still match what the sequencer was invoked with in the beginning. Consequently, after re-reading the todo file the sequencer executes the same first instruction _again_, thus ending up in that "nothing to commit" situation.
Quoted 8 lines
The todo list was never meant to be edited during multi-commit 'git revert' or 'cherry-pick' operations, so perform that "has the todo file been modified" check only when the sequencer was invoked as part of an interactive rebase.
Reported-by Brian Norris <[email protected]>Signed-off-by SZEDER Gábor <[email protected]>---
The patch and the commit message (thank you so much for the detail!) look very good to me.
Quoted 10 lines · Phillip Wood
Quoted 2 lines
Thanks, I suspect it's missing an 'if (
is_rebase_i(opts))' I'll take a look later and come up with a fixThat missing condition was my hunch yesterday evening as well, and while it did fix my tests and didn't break anything else, it took some time to wrap my head around some of the subtleties that are going on in the sequencer. That's why the commit message ended up this long as it did.
Quoted 4 lines
In the end I decided to add the new tests to '
t3429-rebase-edit-todo.sh', because, though these new tests don't actually check 'rebase', that is the one test script focusing on (re-)reading the todo file in particular.
Yup, makes sense.
Thanks,
Dscho
Quoted 61 lines
sequencer.c | 2 +- t/t3429-rebase-edit-todo.sh | 30 ++++++++++++++++++++++++++++++ 2 files changed, 31 insertions(+), 1 deletion(-)sequencer.c +1
-1@@ -3791,7 +3791,7 @@ static int pick_commits(struct repository *r, item->commit, arg, item->arg_len, opts, res, 0); - }else if (check_todo && !res) { + }else if (is_rebase_i (opts) && check_todo && !res) {struct stat st;if (stat (get_todo_path (opts), &st)) {t/t3429-rebase-edit-todo.sh +30
-1@@ -52,4 +52,34 @@ test_expect_success 'todo is re-read after reword and squash' ' test_cmp expected actual ' +test_expect_success 're-reading todo doesnt interfere with revert --edit' ' + git reset --hard third && + + git revert --edit third second && + + cat >expect <<-\EOF && + Revert "second" + Revert "third" + third + second + first + EOF + git log --format="%s" >actual && + test_cmp expect actual +' + +test_expect_success 're-reading todo doesnt interfere with cherry-pick --edit' ' + git reset --hard first && + + git cherry-pick --edit second third && + + cat >expect <<-\EOF && + third + second + first + EOF + git log --format="%s" >actual && + test_cmp expect actual +' + test_done --2.24.0.532.ge18579ded8
Re: [PATCH] sequencer: don't re-read todo for revert and cherry-pick
In reply to SZEDER GáborQuoted 2 lines · SZEDER Gábor
When 'git revert' or 'git cherry-pick --edit' is invoked with multiple commits, then after editing the first commit message is finished both
"commits, then after editing the first commit message, both of" I would say.
Quoted 15 lines
these commands should continue with processing the second commit and launch another editor for its commit message, assuming there are no conflicts, of course.
Alas, this inadvertently changed with commit
a47ba3c777(rebase -i:
check for updated todo after squash and reword, 2019-08-19): after editing the first commit message is finished, both 'git revert' and 'git cherry-pick --edit' exit with error, claiming that "nothing to commit, working tree clean".
...
- When invoking 'git revert' or 'git cherry-pick --edit' with multiple commits they don't read a todo list file but assemble the todo list in memory, thus the associated stat data used to check whether the file has been updated is all zeroed out initially.
Quoted 13 lines
Then the sequencer writes all instructions (including the very first) to the todo file, executes the first 'revert/pick' instruction, and after the user finished editing the commit message the changes of
a47ba3c777kick in, and it checks whether the todo file has been modified. The initial all-zero stat data obviously differs from the todo file's current stat data, so the sequencer concludes that the file has been modified. Technically it is not wrong, of course, because the file just has been written indeed by the sequencer itself, though the file's contents still match what the sequencer was invoked with in the beginning. Consequently, after re-reading the todo file the sequencer executes the same first instruction _again_, thus ending up in that "nothing to commit" situation.
Hmph, that makes it sound as if the right fix is to re-read after writing the first version of the todo file out, so that the stat data matches reality and tells us that it has never been modified?
Quoted 4 lines
The todo list was never meant to be edited during multi-commit 'git revert' or 'cherry-pick' operations, so perform that "has the todo file been modified" check only when the sequencer was invoked as part of an interactive rebase.
OK. That is a valid fix, I think, but the explanation given in the second bullet point gives a wrong impression that it is merely a workaround (iow, we are assuming that the user would behave, instead of making sure we can detect when the user touches the list), when it is *not*.
sequencer.c static int pick_commits(struct repository *r, · line 3794
Quoted 10 lines
sequencer.c +1
-1@@ -3791,7 +3791,7 @@ static int pick_commits(struct repository *r, item->commit, arg, item->arg_len, opts, res, 0); - }else if (check_todo && !res) { + }else if (is_rebase_i (opts) && check_todo && !res) {
It is a bit sad that setting of check_todo is not something a single helper function can decide, so that this is_rebase_i(opts) can be taken into account when that helper function (the logical place would be do_pick_commit()) decides to set (or not set) check_todo.
Unfortunately, that is not sufficient, I suspect. Why did a47ba3c7 ("rebase -i: check for updated todo after squash and reword", 2019-08-19) decide to flip check_todo on when running TODO_EXEC?
Re: [PATCH] sequencer: don't re-read todo for revert and cherry-pick
On 24/11/2019 04:49, Junio C Hamano wrote:
Thanks for working on this Gábor
Quoted 37 lines
Quoted 2 lines · SZEDER Gábor
When 'git revert' or 'git cherry-pick --edit' is invoked with multiple commits, then after editing the first commit message is finished both
"commits, then after editing the first commit message, both of" I would say.
Quoted 15 lines
these commands should continue with processing the second commit and launch another editor for its commit message, assuming there are no conflicts, of course.
Alas, this inadvertently changed with commit
a47ba3c777(rebase -i:
check for updated todo after squash and reword, 2019-08-19): after editing the first commit message is finished, both 'git revert' and 'git cherry-pick --edit' exit with error, claiming that "nothing to commit, working tree clean".
...
- When invoking 'git revert' or 'git cherry-pick --edit' with multiple commits they don't read a todo list file but assemble the todo list in memory, thus the associated stat data used to check whether the file has been updated is all zeroed out initially.Quoted 13 lines
Then the sequencer writes all instructions (including the very first) to the todo file, executes the first 'revert/pick' instruction, and after the user finished editing the commit message the changes of
a47ba3c777kick in, and it checks whether the todo file has been modified. The initial all-zero stat data obviously differs from the todo file's current stat data, so the sequencer concludes that the file has been modified. Technically it is not wrong, of course, because the file just has been written indeed by the sequencer itself, though the file's contents still match what the sequencer was invoked with in the beginning. Consequently, after re-reading the todo file the sequencer executes the same first instruction _again_, thus ending up in that "nothing to commit" situation.
Quoted 3 lines
Hmph, that makes it sound as if the right fix is to re-read after writing the first version of the todo file out, so that the stat data matches reality and tells us that it has never been modified?
I think we should update the stat data after we write the todo list. ag/sequencer-todo-updates changes the rebase startup sequence to avoid the initial read that populates the stat data. There's a subtle difference in the todo list handling between rabese and cherry-pick/revert. The list written by rebase does not include the commit currently being picked but it does for cherry-pick/revert. This means that rebase is not subject to this bug in ag/sequencer-todo-updates as although it re-reads the todo list because the stat data is zero the list does not contain the commit we've just picked.
Quoted 6 lines
Quoted 5 lines
The todo list was never meant to be edited during multi-commit 'git revert' or 'cherry-pick' operations, so perform that "has the todo file been modified" check only when the sequencer was invoked as part of an interactive rebase.
Quoted 6 lines
OK. That is a valid fix, I think, but the explanation given in the second bullet point gives a wrong impression that it is merely a workaround (iow, we are assuming that the user would behave, instead of making sure we can detect when the user touches the list), when it is *not*.
I'd be happier not to set check_todo in the first place unless we're rebasing (though one could argue that Gábor's version is more future-proof)
Quoted 16 lines
sequencer.c static int pick_commits(struct repository *r, · line 3794
Quoted 10 lines
sequencer.c +1
-1@@ -3791,7 +3791,7 @@ static int pick_commits(struct repository *r, item->commit, arg, item->arg_len, opts, res, 0); - }else if (check_todo && !res) { + }else if (is_rebase_i (opts) && check_todo && !res) {It is a bit sad that setting of
check_todois not something a single helper function can decide, so that thisis_rebase_i(opts)can be taken into account when that helper function (the logical place would bedo_pick_commit()) decides to set (or not set)check_todo.
Quoted 3 lines
Unfortunately, that is not sufficient, I suspect. Why did
a47ba3c7("rebase -i: check for updated todo after squash and reword", 2019-08-19) decide to flipcheck_todoon when runningTODO_EXEC?
I'm not sure what you mean by this
This is what I had before I saw Gábor's patch (the tests are pretty similar but I think we should check that the messages of all the picks are actually edited with --edit - that does not seem to be checked by the current tests) Note this does not include updating the stat data after writing the todo list yet as I've only just thought about that. Gábor are you happy to take this forward?
Best Wishes
Phillip
--- 8> ---
sequencer.c +1 -1
@@ -1999,7 +1999,7 @@ static int do_pick_commit(struct repository *r, res =do_commit (r, msg_file, author, opts, flags); else res =error (_ ("unable to parse commit author" )); - *check_todo = !!(flags &EDIT_MSG ); + *check_todo =is_rebase_i (opts) && !!(flags &EDIT_MSG );if (!res && reword) { fast_forward_edit: res = run_git_commit(r, NULL, opts, EDIT_MSG |
t/t3508-cherry-pick-many-commits.sh +43 -0
52 lines not on this page · raw message
'
+test_expect_success 'revert --edit first..fourth works' '
+ git checkout -f master && + git reset --hard fourth && + test_tick && + GIT_EDITOR="echo edited >>" git revert --edit first..fourth && + git log --pretty=format:%B -3 >actual && + cat >expect <<-EOF && + Revert "second"
+
+ This reverts commit $(git rev-parse second).
+
+ edited
+
+ Revert "third"
+
+ This reverts commit $(git rev-parse third).
+
+ edited
+
+ Revert "fourth"
+
+ This reverts commit $(git rev-parse fourth).
+
+ edited
+ EOF
+ test_cmp expect actual &&
+ git diff --quiet first &&
+ git diff --cached --quiet first &&
+ git diff --quiet HEAD first
+'
+
test_expect_success 'revert ^first fourth works' '
git checkout -f master &&
git reset --hard fourth &&PATCH 1/1 t3429: try to protect against a potential racy todo file problem
Quoted 41 lines · Phillip Wood
On 24/11/2019 04:49, Junio C Hamano wrote:
Thanks for working on this Gábor
Quoted 37 lines
Quoted 2 lines · SZEDER Gábor
When 'git revert' or 'git cherry-pick --edit' is invoked with multiple commits, then after editing the first commit message is finished both
"commits, then after editing the first commit message, both of" I would say.
Quoted 28 lines
these commands should continue with processing the second commit and launch another editor for its commit message, assuming there are no conflicts, of course.
Alas, this inadvertently changed with commit
a47ba3c777(rebase -i:
check for updated todo after squash and reword, 2019-08-19): after editing the first commit message is finished, both 'git revert' and 'git cherry-pick --edit' exit with error, claiming that "nothing to commit, working tree clean".
...
- When invoking 'git revert' or 'git cherry-pick --edit' with multiple commits they don't read a todo list file but assemble the todo list in memory, thus the associated stat data used to check whether the file has been updated is all zeroed out initially.Then the sequencer writes all instructions (including the very first) to the todo file, executes the first 'revert/pick' instruction, and after the user finished editing the commit message the changes of
a47ba3c777kick in, and it checks whether the todo file has been modified. The initial all-zero stat data obviously differs from the todo file's current stat data, so the sequencer concludes that the file has been modified. Technically it is not wrong, of course, because the file just has been written indeed by the sequencer itself, though the file's contents still match what the sequencer was invoked with in the beginning. Consequently, after re-reading the todo file the sequencer executes the same first instruction _again_, thus ending up in that "nothing to commit" situation.
Quoted 5 lines
Quoted 3 lines
Hmph, that makes it sound as if the right fix is to re-read after writing the first version of the todo file out, so that the stat data matches reality and tells us that it has never been modified?
I think we should update the stat data after we write the todo list.
Well, yes and no.
No, because we are dealing with regression in v2.24.0 here, so the simpler the fix the better it is for maint. I don't think a fix can get any simpler than my patch, with or without the suggestions from Phillip.
Yes, we should definitely consider updating the stat data after the sequencer writes the todo list, or any other options with which the sequencer could notice a modified todo list file with less subtlety. Alas, there is a big can of worms in that direction, see the patch below, and we have to be very careful going that way, so I think it's only viable in the long term, but less suitable as a regression fix for maint.
(Hrm, perhaps I spent too many words on the all zeroed out stat data, and managed to sidetrack you a bit...)
--- >8 ---
Subject: [PATCH] t3429: try to protect against a potential racy todo file
problem
During an interactive rebase session the user can run basically any commands with an 'exec' instruction, including commands that modify the todo list of the ongoing rebase. Similarly, the user might choose to modify the todo list while editing a commit message for a 'reword' or a 'squash' instruction. The ongoing interactive rebase process should be able to notice the modified todo file and re-read it to follow the updated instructions. This has been working well for the 'exec' instruction for some time, but for 'reword' and 'squash' it has been fixed only recently in a47ba3c777 (rebase -i: check for updated todo after squash and reword, 2019-08-19).
To notice a changed todo file the sequencer stores the file's stat data in its 'struct todo_list' instance, and compares it with the file's current stat data after 'reword', 'squash' and 'exec' instructions. If the two stat data doesn't match, it re-reads the todo file.
Sounds simple, but there are some subtleties going on here:
- The 'struct todo_list' holds the stat data from the time when the todo file was last read.
- This stat data in 'struct todo_list' is not updated when the sequencer itself writes the todo file.
- Before executing each instruction during an interactive rebase, the sequencer always updates the todo file by removing the just-about-to-be-executed instruction. This changes the file's size and inode [1].
Consequently, when the sequencer looks at the stat data after a 'reword', 'squash' or 'exec' instruction, it most likely finds that they differ, even when the user didn't modify the todo list at all! This is not an issue in practice, it just wastes a few cycles on re-reading the todo list that matches what the sequencer already has in memory anyway.
However, an unsuspecting Git developer might try to "fix" it by simply updating the stat data each time the sequencer writes the todo list for an interactive rebase. On first sight it looks quite sensible and straightforward, but we have to be very careful when doing that, because potential racy problems lie that way.
It is possible to overwrite the todo list file without modifying either its inode or size, and if such an overwrite were to happen in the same second when the file was last read (our stat data has one second granularity by default), then the actual stat data on the file system would match the stat data that the sequencer has in memory. Consequently, such a modification to the todo list file would go unnoticed.
Add a test to 't3429-rebase-edit-todo.sh' that modifies the todo list during an interactive rebase in a way that the file's stat data very likely doesn't change. Since we have no control over ctime repeat the test a few times, just like we do in the racy git and racy split index tests (t0010 and t1701). This test succeeds right now because of all the above, but it does fail most of the time if the stat data is naively updated when writing the todo file [2]. Hopefully this test will help prevent us from introducing such racy problems if we ever change how the sequencer looks out for modified todo list files.
This new test only checks what it's supposed to when it doesn't change the size of the todo file while modifying it, thus it critically depends on the todo file format. And the todo file format did change in the past: early on it stored abbreviated commit object ids, but since 75c6976655 (rebase -i: fix short SHA-1 collision, 2013-08-23) it stores full object ids. Perhaps it will change in the future as well, so be extra defensive about this, and check that the file size before and after the edit actually match, and complain with a "bug in the test script" if they don't.
[1] The todo file is written using the lockfile API, i.e. first to the lockfile, which is then moved to place, so the new file can't possibly have the same inode as the file it replaces. Note, however, that the file system might reuse inodes, so it is possible that the new todo file ends up with the same inode as is recorded in the 'struct todo_list' from the last time the file was read.
[2] To trigger failure in the new test updating the stat data when
writing the todo file like the diff below does. No other test in
our test suite seems to fail with this change applied.
diff --git a/sequencer.c b/sequencer.c
index 3b05d0277d..b2acb7d536 100644
--- a/sequencer.c
+++ b/sequencer.c
@@ -2730,6 +2730,7 @@ static int save_todo(struct todo_list *todo_list, struct replay_opts *opts)
struct lock_file todo_lock = LOCK_INIT;
const char *todo_path = get_todo_path(opts);
int next = todo_list->current, offset, fd;
+ struct stat st;
/*
* rebase -i writes "git-rebase-todo" without the currently executing
@@ -2748,6 +2749,10 @@ static int save_todo(struct todo_list *todo_list, struct replay_opts *opts)
if (commit_lock_file(&todo_lock) < 0)
return error(_("failed to finalize '%s'"), todo_path);
+ if (stat(todo_path, &st) < 0)
+ return error(_("could not stat '%s'"), todo_path);
+ fill_stat_data(&todo_list->stat, &st);
+
if (is_rebase_i(opts) && next > 0) {
const char *done = rebase_path_done();
int fd = open(done, O_CREAT | O_WRONLY | O_APPEND, 0666);--- t/t3429-rebase-edit-todo.sh | 59 +++++++++++++++++++++++++++++++++++++ 1 file changed, 59 insertions(+)
t/t3429-rebase-edit-todo.sh +59 -0
@@ -52,4 +52,63 @@ test_expect_success 'todo is re-read after reword and squash' ' test_cmp expected actual ' +test_expect_success 'setup for racy tests' ' + write_script sequence-editor <<-EOS && + cat >.git/rebase-merge/git-rebase-todo <<-\EOF + r $(git rev-parse second^0) second + r $(git rev-parse third^0) third + EOF + EOS + + write_script commit-editor <<-\EOS && + read first_line <"$1" && + echo "$first_line - edited" >"$1" && + + todo=.git/rebase-merge/git-rebase-todo && + + if test "$first_line" = second + then + old_size=$(wc -c <"$todo") && + # Replace the "reword <full-oid> third" line with + # a different instruction of the same line length. + # Overwrite the file in-place, so the inode stays + # the same as well. + cat >"$todo" <<-EOF && + exec echo 0123456789012345678901234 >exec-cmd-was-run + EOF + new_size=$(wc -c <"$todo") && + + if test $old_size -ne $new_size + then + echo >&2 "error: bug in the test script: the size of the todo list must not change" + exit 1 + fi + fi + EOS + + cat >expect <<-\EOF + second - edited + first + EOF +' + +# This test can give false success if your machine is sufficiently +# slow or all trials happened to happen on second boundaries. + +for trial in 0 1 2 3 4 +do + test_expect_success "racy todo re-read #$trial" ' + git reset --hard third && + rm -rf exec-cmd-was-run && + + GIT_SEQUENCE_EDITOR=./sequence-editor \ + GIT_EDITOR=./commit-editor \ + git rebase -i HEAD^^ && + + test_path_is_file exec-cmd-was-run && + git log --format=%s >actual && + test_cmp expect actual + ' +done + test_done
Signature · 1 line
2.24.0.532.ge18579ded8
Re: [PATCH] t3429: try to protect against a potential racy todo file problem
Quoted 7 lines · SZEDER Gábor
Quoted 6 lines · Phillip Wood
On 24/11/2019 04:49, Junio C Hamano wrote:
...Quoted 3 lines
Hmph, that makes it sound as if the right fix is to re-read after writing the first version of the todo file out, so that the stat data matches reality and tells us that it has never been modified?
Quoted 8 lines
Quoted 1 line
I think we should update the stat data after we write the todo list.
Well, yes and no.
No, because we are dealing with regression in v2.24.0 here, so the simpler the fix the better it is for maint. I don't think a fix can get any simpler than my patch, with or without the suggestions from Phillip.
Of course, the simplest "fix" for regression is to revert the offending one, and anything else is a band-aid ;-). The question is which band-aid is the least risky and which one takes us the closest to the real solution. I tend to agree that forcing to skip checking no matter what the variable "check_todo" says unless is_rebase_i() qualifies as the band-aid that is the least risky.
Quoted 7 lines
Yes, we should definitely consider updating the stat data after the sequencer writes the todo list, or any other options with which the sequencer could notice a modified todo list file with less subtlety. Alas, there is a big can of worms in that direction, see the patch below, and we have to be very careful going that way, so I think it's only viable in the long term, but less suitable as a regression fix for maint.
Yes, I agree that it is much less suitable than even reverting the offending one outright.
Quoted 2 lines
(Hrm, perhaps I spent too many words on the all zeroed out stat data, and managed to sidetrack you a bit...)
No, I do not think so. Thinking about what we need to do in the longer term, while coming up with a shorter term fix, is a necessary step of gaining confidence in the latter.
Again, thanks both for thinking about this issue.
Re: [PATCH] t3429: try to protect against a potential racy todo file problem
In reply to SZEDER GáborQuoted 40 lines · SZEDER Gábor
To notice a changed todo file the sequencer stores the file's stat data in its 'struct
todo_list' instance, and compares it with the file's current stat data after 'reword', 'squash' and 'exec' instructions. If the two stat data doesn't match, it re-reads the todo file.Sounds simple, but there are some subtleties going on here:
- The 'struct
todo_list' holds the stat data from the time when the todo file was last read.- This stat data in 'struct
todo_list' is not updated when the sequencer itself writes the todo file.- Before executing each instruction during an interactive rebase, the sequencer always updates the todo file by removing the just-about-to-be-executed instruction. This changes the file's size and inode [1].
Consequently, when the sequencer looks at the stat data after a 'reword', 'squash' or 'exec' instruction, it most likely finds that they differ, even when the user didn't modify the todo list at all! This is not an issue in practice, it just wastes a few cycles on re-reading the todo list that matches what the sequencer already has in memory anyway.
However, an unsuspecting Git developer might try to "fix" it by simply updating the stat data each time the sequencer writes the todo list for an interactive rebase. On first sight it looks quite sensible and straightforward, but we have to be very careful when doing that, because potential racy problems lie that way.
It is possible to overwrite the todo list file without modifying either its inode or size, and if such an overwrite were to happen in the same second when the file was last read (our stat data has one second granularity by default), then the actual stat data on the file system would match the stat data that the sequencer has in memory. Consequently, such a modification to the todo list file would go unnoticed.
Quoted 7 lines
[1] The todo file is written using the lockfile API, i.e. first to the lockfile, which is then moved to place, so the new file can't possibly have the same inode as the file it replaces. Note, however, that the file system might reuse inodes, so it is possible that the new todo file ends up with the same inode as is recorded in the 'struct
todo_list' from the last time the file was read.
Unfortunately, we already do have an issue here when the sequencer can overlook a modified todo list, but triggering it needs the lucky "alignment" of inodes as well. I can trigger it fairly reliably with the test below, but forcing the inode match makes it kind of gross and Linux-only.
https://travis-ci.org/szeder/git/jobs/616460522#L1470
--- >8 ---
t/t9999-rebase-racy-todo-reread.sh new file +87 -0
@@ -0,0 +1,87 @@ +#!/bin/sh + +test_description='racy edit todo reread problem' + +. ./test-lib.sh + +test_expect_success 'setup' ' + test_commit first_ && + test_commit second && + test_commit third_ && + test_commit fourth && + test_commit fifth_ && + test_commit sixth_ && + + write_script sequence-editor <<-\EOS && + todo=.git/rebase-merge/git-rebase-todo && + cat >"$todo" <<-EOF + reword $(git rev-parse second^0) second + reword $(git rev-parse third_^0) third_ + reword $(git rev-parse fourth^0) fourth + EOF + EOS + + write_script commit-editor <<-\EOS && + read first_line <"$1" && + echo "$first_line - edited" >"$1" && + + todo=.git/rebase-merge/git-rebase-todo && + + if test "$first_line" = second + then + stat --format=%i "$todo" >expected-ino + elif test "$first_line" = third_ + then + ino=$(cat expected-ino) && + file=$(find . -inum $ino) && + if test -n "$file" + then + echo && + echo "Trying to free inode $ino by moving \"$file\" out of the way" && + cp -av "$file" "$file".tmp && + rm -fv "$file" + fi && + + cat >"$todo".tmp <<-EOF && + reword $(git rev-parse fifth_^0) fifth_ + reword $(git rev-parse sixth_^0) sixth_ + EOF + mv -v "$todo".tmp "$todo" && + + if test "$ino" -eq $(stat --format=%i "$todo") + then + echo "Yay! The todo list did get inode $ino, just what the sequencer is expecting!" + fi && + + if test -n "$file" + then + mv -v "$file".tmp "$file" + fi + fi + EOS + + cat >expect <<-\EOF + sixth_ - edited + fifth_ - edited + third_ - edited + second - edited + first_ + EOF +' + +for trial in 0 1 2 3 4 +do + test_expect_success "demonstrate racy todo re-read problem #$trial" ' + git reset --hard fourth && + >expected-ino && # placeholder + + GIT_SEQUENCE_EDITOR=./sequence-editor \ + GIT_EDITOR=./commit-editor \ + git rebase -i HEAD^^^ && + + git log --format=%s >actual && + test_cmp expect actual + ' +done + +test_done
Re: [PATCH] t3429: try to protect against a potential racy todo file problem
In reply to SZEDER GáborQuoted 19 lines · SZEDER Gábor
To notice a changed todo file the sequencer stores the file's stat data in its 'struct
todo_list' instance, and compares it with the file's current stat data after 'reword', 'squash' and 'exec' instructions. If the two stat data doesn't match, it re-reads the todo file.Sounds simple, but there are some subtleties going on here:
- The 'struct
todo_list' holds the stat data from the time when the todo file was last read.- This stat data in 'struct
todo_list' is not updated when the sequencer itself writes the todo file.- Before executing each instruction during an interactive rebase, the sequencer always updates the todo file by removing the just-about-to-be-executed instruction. This changes the file's size and inode [1].
Quoted 6 lines
Consequently, when the sequencer looks at the stat data after a 'reword', 'squash' or 'exec' instruction, it most likely finds that they differ, even when the user didn't modify the todo list at all! This is not an issue in practice, it just wastes a few cycles on re-reading the todo list that matches what the sequencer already has in memory anyway.
It can be much more than just a few cycles, because the total number of parsed instructions from all the todo file re-reads can go quadratic with the number of rebased commits.
The simple test below runs 'git rebase -i -x' on 1000 commits, which takes over 14seconds to run. If it doesn't re-read the todo file at all (I simply deleted the whole condition block checking the stat data and re-reading) it runs for only ~2.5secs.
Just another angle to consider...
--- >8 ---
test_expect_success 'test' '
num_commits=1000 && test_commit_bulk --filename=file $num_commits && /usr/bin/time -f "Elapsed time: %e" \ git rebase -i --root -x true 2>out && grep "Executing: true" out >actual && test_line_count = $num_commits actual && # show the elapsed time tail -n2 out
'
--- >8 ---
Re: [PATCH] t3429: try to protect against a potential racy todo file problem
Quoted 31 lines · SZEDER Gábor
Quoted 19 lines · SZEDER Gábor
To notice a changed todo file the sequencer stores the file's stat data in its 'struct
todo_list' instance, and compares it with the file's current stat data after 'reword', 'squash' and 'exec' instructions. If the two stat data doesn't match, it re-reads the todo file.Sounds simple, but there are some subtleties going on here:
- The 'struct
todo_list' holds the stat data from the time when the todo file was last read.- This stat data in 'struct
todo_list' is not updated when the sequencer itself writes the todo file.- Before executing each instruction during an interactive rebase, the sequencer always updates the todo file by removing the just-about-to-be-executed instruction. This changes the file's size and inode [1].
Quoted 6 lines
Consequently, when the sequencer looks at the stat data after a 'reword', 'squash' or 'exec' instruction, it most likely finds that they differ, even when the user didn't modify the todo list at all! This is not an issue in practice, it just wastes a few cycles on re-reading the todo list that matches what the sequencer already has in memory anyway.
It can be much more than just a few cycles, because the total number of parsed instructions from all the todo file re-reads can go quadratic with the number of rebased commits.
Quoted 6 lines
The simple test below runs 'git rebase -i -x' on 1000 commits, which takes over 14seconds to run. If it doesn't re-read the todo file at all (I simply deleted the whole condition block checking the stat data and re-reading) it runs for only ~2.5secs.
Just another angle to consider...
I know dscho was keen to avoid re-parsing the list all the time [1] presumably because of the quadratic behavior. (He also assumed most people were using ns stat times [2] but that appears not to be the case) Could we just compare the text of the todo list on disk to whats in todo->buf.buf (with an appropriate offset)? That would avoid parsing the text and looking up all the commits with get_oid()
Best Wishes
Phillip
[1] https://public-inbox.org/git/alpine.DEB.2.20.1703021617510.3767@virtualbox/
[2] https://public-inbox.org/git/alpine.DEB.2.20.1704131526500.2135@virtualbox/
Quoted 19 lines
--- >8 ---
test_expect_success'test' 'num_commits=1000 && test_commit_bulk --filename=file $num_commits && /usr/bin/time -f "Elapsed time: %e" \ git rebase -i --root -x true 2>out && grep "Executing: true" out >actual && test_line_count = $num_commits actual && # show the elapsed time tail -n2 out'
--- >8 ---
Re: [PATCH] t3429: try to protect against a potential racy todo file problem
Quoted 37 lines · Phillip Wood
Quoted 36 lines · SZEDER Gábor
Quoted 19 lines · SZEDER Gábor
To notice a changed todo file the sequencer stores the file's stat data in its 'struct
todo_list' instance, and compares it with the file's current stat data after 'reword', 'squash' and 'exec' instructions. If the two stat data doesn't match, it re-reads the todo file.Sounds simple, but there are some subtleties going on here:
- The 'struct
todo_list' holds the stat data from the time when the todo file was last read.- This stat data in 'struct
todo_list' is not updated when the sequencer itself writes the todo file.- Before executing each instruction during an interactive rebase, the sequencer always updates the todo file by removing the just-about-to-be-executed instruction. This changes the file's size and inode [1].
Quoted 6 lines
Consequently, when the sequencer looks at the stat data after a 'reword', 'squash' or 'exec' instruction, it most likely finds that they differ, even when the user didn't modify the todo list at all! This is not an issue in practice, it just wastes a few cycles on re-reading the todo list that matches what the sequencer already has in memory anyway.
It can be much more than just a few cycles, because the total number of parsed instructions from all the todo file re-reads can go quadratic with the number of rebased commits.
The simple test below runs 'git rebase -i -x' on 1000 commits, which takes over 14seconds to run. If it doesn't re-read the todo file at all (I simply deleted the whole condition block checking the stat data and re-reading) it runs for only ~2.5secs.
Quoted 5 lines
Quoted 1 line
Just another angle to consider...
I know dscho was keen to avoid re-parsing the list all the time [1] presumably because of the quadratic behavior. (He also assumed most people were using ns stat times [2] but that appears not to be the case)
Nanosecond file timestamp comparisons are only enabled by the USE_NSEC macro, which is only defined if the USE_NSEC Makefile knob is enabled, but that is not enabled by default.
Then there is the related NO_NSEC Makefile knob:
# Define NO_NSEC if your "struct stat" does not have "st_ctim.tv_nsec"
# available. This automatically turns USE_NSEC off.
As Dscho mentioned in [2], we do disable nanosecond file timestamp comparisons in 'config.mak.uname' on a handful of platforms by setting NO_NSEC. This, however, does not mean that nanosec timestamps are enabled on other platforms by default.
Quoted 4 lines
Could we
just compare the text of the todo list on disk to whats intodo->buf.buf(with an appropriate offset)? That would avoid parsing the text and looking up all the commits withget_oid()
Comparing the contents without parsing is still quadratic in the size of the todo list, though I suppose with a much lower constant factor than actually parsing it.
Quoted 24 lines
[1]
https://public-inbox.org/git/alpine.DEB.2.20.1703021617510.3767@virtualbox/
[2]
https://public-inbox.org/git/alpine.DEB.2.20.1704131526500.2135@virtualbox/Quoted 19 lines
--- >8 ---
test_expect_success'test' 'num_commits=1000 && test_commit_bulk --filename=file $num_commits && /usr/bin/time -f "Elapsed time: %e" \ git rebase -i --root -x true 2>out && grep "Executing: true" out >actual && test_line_count = $num_commits actual && # show the elapsed time tail -n2 out'
--- >8 ---
Re: [PATCH] t3429: try to protect against a potential racy todo file problem
Quoted 58 lines · SZEDER Gábor
Quoted 37 lines · Phillip Wood
Quoted 36 lines · SZEDER Gábor
Quoted 19 lines · SZEDER Gábor
To notice a changed todo file the sequencer stores the file's stat data in its 'struct
todo_list' instance, and compares it with the file's current stat data after 'reword', 'squash' and 'exec' instructions. If the two stat data doesn't match, it re-reads the todo file.Sounds simple, but there are some subtleties going on here:
- The 'struct
todo_list' holds the stat data from the time when the todo file was last read.- This stat data in 'struct
todo_list' is not updated when the sequencer itself writes the todo file.- Before executing each instruction during an interactive rebase, the sequencer always updates the todo file by removing the just-about-to-be-executed instruction. This changes the file's size and inode [1].
Quoted 6 lines
Consequently, when the sequencer looks at the stat data after a 'reword', 'squash' or 'exec' instruction, it most likely finds that they differ, even when the user didn't modify the todo list at all! This is not an issue in practice, it just wastes a few cycles on re-reading the todo list that matches what the sequencer already has in memory anyway.
It can be much more than just a few cycles, because the total number of parsed instructions from all the todo file re-reads can go quadratic with the number of rebased commits.
The simple test below runs 'git rebase -i -x' on 1000 commits, which takes over 14seconds to run. If it doesn't re-read the todo file at all (I simply deleted the whole condition block checking the stat data and re-reading) it runs for only ~2.5secs.
Quoted 5 lines
Quoted 1 line
Just another angle to consider...
I know dscho was keen to avoid re-parsing the list all the time [1] presumably because of the quadratic behavior. (He also assumed most people were using ns stat times [2] but that appears not to be the case)
Nanosecond file timestamp comparisons are only enabled by the
USE_NSECmacro, which is only defined if theUSE_NSECMakefileknob is enabled, but that is not enabled by default.Then there is the related
NO_NSECMakefileknob:# Define
NO_NSECif your "struct stat" does not have "st_ctim.tv_nsec"# available. This automatically turns USE_NSEC off.As Dscho mentioned in [2], we do disable nanosecond file timestamp comparisons in 'config.mak.uname' on a handful of platforms by setting
NO_NSEC. This, however, does not mean that nanosec timestamps are enabled on other platforms by default.
Quoted 8 lines
Quoted 4 lines
Could we
just compare the text of the todo list on disk to whats intodo->buf.buf(with an appropriate offset)? That would avoid parsing the text and looking up all the commits withget_oid()Comparing the contents without parsing is still quadratic in the size of the todo list, though I suppose with a much lower constant factor than actually parsing it.
The patch below (assuming thunderbird doesn't mangle it) reduces the time to run your bulk commit test from 30s to 7s, if I delete the condition block which checks the stat data it takes 4.7s on my (somewhat ancient) laptop. So there is a cost to the string comparison approach but it's much less that the full todo list parsing.
Best Wishes
Phillip
--- >8 ---
sequencer.c +10 -5
@@ -3909,12 +3909,17 @@ static int pick_commits(struct repository *r, arg, item->arg_len, opts, res, 0); }else if (check_todo && !res) { -struct stat st; - -if (stat (get_todo_path (opts), &st)) { - res =error_errno (_ ("could not stat '%s'" ), +int offset; +struct strbuf buf =STRBUF_INIT ; +if (strbuf_read_file (&buf,get_todo_path (opts),8096 ) <0 ) + res =error_errno (_ ("could not read '%s'" ),get_todo_path (opts)); - }else if (match_stat_data (&todo_list->stat, &st)) { + + offset =get_item_line_offset (todo_list, + todo_list->current +1 ); +if (buf.len != todo_list->buf.len - offset || +memcmp (buf.buf, todo_list->buf.buf + offset, buf.len)) { +fputs ("re-reading todo list\n" , stderr);/* Reread the todo file if it has
changed. */
todo_list_release(todo_list); if (read_populate_todo(r, todo_list, opts))
Re: [PATCH] sequencer: don't re-read todo for revert and cherry-pick
In reply to Phillip WoodQuoted 12 lines · Phillip Wood
Quoted 11 lines
Review
Quoted 2 lines
(quoted diff) +1
-1- } else if (check_todo && !res) { + } else if (is_rebase_i(opts) && check_todo && !res) {It is a bit sad that setting of
check_todois not something a single helper function can decide, so that thisis_rebase_i(opts)can be taken into account when that helper function (the logical place would bedo_pick_commit()) decides to set (or not set)check_todo.Unfortunately, that is not sufficient, I suspect. Why did
a47ba3c7("rebase -i: check for updated todo after squash and reword", 2019-08-19) decide to flipcheck_todoon when runningTODO_EXEC?
Quoted 6 lines
I'm not sure what you mean by this
This is what I had before I saw Gábor's patch (the tests are pretty similar but I think we should check that the messages of all the picks are actually edited with --edit - that does not seem to be checked by the current tests)...
I first thought that unsetting *check_todo in do_pick_commit(), when !is_rebase_i(), was a clean solution. But sadly it is *not* a godo equivalent to Gábor's patch, because check_todo can be set to true without looking at is_rebase_i() in pick_commits() [*1*]. To ignore that setting where the variable's value is used, the hunk we see above in the beginning of this message is necessary.
That was what I meant. I think the "This is what I had before" patch matches my "I first thought" version, so we were on the same page and both wrong ;-)
Thanks.
[Footnote]
*1* I still do not know why a47ba3c7 ("rebase -i: check for updated todo after squash and reword", 2019-08-19) sets check_todo to true without looking at is_rebase_i(). If this unconditonal setting in TODO_EXEC did not exist, I think your "This is what I had before" patch would have been equivalent to Gábor's patch.
Re: [PATCH] sequencer: don't re-read todo for revert and cherry-pick
Hi Junio
Quoted 21 lines · Junio C Hamano
Quoted 12 lines · Phillip Wood
Quoted 11 lines
Review
Quoted 2 lines
(quoted diff) +1
-1- } else if (check_todo && !res) { + } else if (is_rebase_i(opts) && check_todo && !res) {It is a bit sad that setting of
check_todois not something a single helper function can decide, so that thisis_rebase_i(opts)can be taken into account when that helper function (the logical place would bedo_pick_commit()) decides to set (or not set)check_todo.Unfortunately, that is not sufficient, I suspect. Why did
a47ba3c7("rebase -i: check for updated todo after squash and reword", 2019-08-19) decide to flipcheck_todoon when runningTODO_EXEC?Quoted 6 lines
I'm not sure what you mean by this
This is what I had before I saw Gábor's patch (the tests are pretty similar but I think we should check that the messages of all the picks are actually edited with --edit - that does not seem to be checked by the current tests)...
Quoted 4 lines
I first thought that unsetting
*check_todoindo_pick_commit(), when !is_rebase_i(), was a clean solution. But sadly it is *not* a godo equivalent to Gábor's patch, becausecheck_todocan be set to true without looking atis_rebase_i()inpick_commits()[*1*].
That only happens if we're running a rebase as exec commands are not used by cherry-pick and revert so the unconditional setting that remains after my suggested fix should be safe. Prior to a47ba3c7 ("rebase -i: check for updated todo after squash and reword", 2019-08-19) we always checked for an updated todo list after processing an exec command without checking is_rebase_i()
Best Wishes
Phillip
Quoted 18 lines
To ignore
that setting where the variable's value is used, the hunk we see above in the beginning of this message is necessary.That was what I meant. I think the "This is what I had before" patch matches my "I first thought" version, so we were on the same page and both wrong ;-)
Thanks.
[Footnote]
*1* I still do not know why
a47ba3c7("rebase -i: check for updated todo after squash and reword", 2019-08-19) setscheck_todoto true without looking atis_rebase_i(). If this unconditonal setting inTODO_EXECdid not exist, I think your "This is what I had before" patch would have been equivalent to Gábor's patch.