git rebase --skip stuck in a loop

git rebase --skip stuck in a loop

[Phillip Susi]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I'm in the middle of a long rebase and have had no trouble with git rebase --skip several times, but now it has become stuck:

psusi@devserv:~/parted.git$ git rebase --skip
Auto-merging libparted/arch/linux.c
CONFLICT (content): Merge conflict in libparted/arch/linux.c

When you have resolved this problem, run "git rebase --continue".
If you prefer to skip this patch, run "git rebase --skip" instead.
To check out the original branch and stop rebasing, run "git rebase --abort".

psusi@devserv:~/parted.git$ cat .git/rebase-merge/msgnum
17

Each time I try to skip, it just keeps trying to reapply this one patch.  Any ideas?

git version 1.9.1

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.17 (MingW32)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJTl1UcAAoJEI5FoCIzSKrwwHIH/38Cm19zg+37zgckLiy/3GhN
3Gmil5kX+3KkIHCxlPz3Ti3xCA5baM7tDzFdDIKcx8N/i8oALgWeAWf1Euy9Ww1K
3etIAMKzO463kmV7UbgSbLz5DpYWSaGo9WiYAC7xklhQV94w1Ainx5Lo4kRv1Wfj
R9TpQgViFnW2gNJv1zw0qHLXk1/h88LlAQsBaaY6I4f/DOLhAte7rGinkTgZtjmo
G/9PUMudQcehG65ITPlNLtoFsM8UHadNMLwJts/B7Yq23XNyRF50IaT8c1A/irSU
mfYqdHCho3D4kq+k0u+t0Z0bj6pfvo4b0khLafrYLrWGHC5K+Z3lE63ysJ/Mdj8=
=LZ9q
-----END PGP SIGNATURE-----

Re: git rebase --skip stuck in a loop

[Phillip Susi]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

So nobody has any ideas on what to check for or how to debug this?

On 6/10/2014 2:57 PM, Phillip Susi wrote:
> I'm in the middle of a long rebase and have had no trouble with
> git rebase --skip several times, but now it has become stuck:
> 
> psusi@devserv:~/parted.git$ git rebase --skip Auto-merging
> libparted/arch/linux.c CONFLICT (content): Merge conflict in
> libparted/arch/linux.c
> 
> When you have resolved this problem, run "git rebase --continue". 
> If you prefer to skip this patch, run "git rebase --skip" instead. 
> To check out the original branch and stop rebasing, run "git
> rebase --abort".
> 
> psusi@devserv:~/parted.git$ cat .git/rebase-merge/msgnum 17
> 
> Each time I try to skip, it just keeps trying to reapply this one 
> patch.  Any ideas?
> 
> git version 1.9.1
> 
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.17 (MingW32)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJTmhUcAAoJEI5FoCIzSKrwkjgIAKOclhuJPiNoWIEv1dBr4DBC
IdwG9hY8lQPCN5Pg8th5CYk3ziX7iZ8+jaHEBaUYX2yehT1deg5WfsxU0uezWphH
JyHMRX4kk7l1PW3/v3bEvZ0WYe77s4GB3m9XegjKwEL8xtGi7srEPsHgWB8gnFzE
hswUMbq5mw9hoIpYnxEs18F2MOfP6i4J3gTilPrmq+hZCQyZrX/IsV5lR6kDXRES
j7b3cr6n2EfUeWxKrwo+tMIBdGAgpMamWlzqM7gMND/YUswv84mD3b9lXjEfjqZf
GfBXJSH/z0KLDKycYrDOZlryLEnx///N6STg2WGm0oo7ehAKn6Mtgi1rR5y/aYs=
=bUQV
-----END PGP SIGNATURE-----

Re: git rebase --skip stuck in a loop

[brian m. carlson]

[-- Attachment #1: Type: text/plain, Size: 604 bytes --]

On Thu, Jun 12, 2014 at 05:01:16PM -0400, Phillip Susi wrote:
> So nobody has any ideas on what to check for or how to debug this?

I'm assuming this works in the general case, because we have tests for
it.  Do you perhaps have a repository and a set of reproduction steps we
could use to try to reproduce this?

If you can get that information, I'm interested in trying to fix it.

-- 
brian m. carlson / brian with sandals: Houston, Texas, US
+1 832 623 2791 | http://www.crustytoothpaste.net/~bmc | My opinion only
OpenPGP: RSA v4 4096b: 88AC E9B2 9196 305B A994 7552 F1BA 225C 0223 B187

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: git rebase --skip stuck in a loop

[Phillip Susi]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

On 06/12/2014 09:02 PM, brian m. carlson wrote:
> On Thu, Jun 12, 2014 at 05:01:16PM -0400, Phillip Susi wrote:
>> So nobody has any ideas on what to check for or how to debug
>> this?
> 
> I'm assuming this works in the general case, because we have tests
> for it.  Do you perhaps have a repository and a set of reproduction
> steps we could use to try to reproduce this?
> 
> If you can get that information, I'm interested in trying to fix
> it.

I don't have a set of reproduction steps, but I do still have the repo
that is in this stuck state if there is any data you might want me to
pull out of it ( .git/rebase-apply? ).

I had spoken to another Ubuntu dev who said that he had this happen to
him once too and he just did an abort and tried again and the problem
went away, so it seems hard to reproduce, which is why I have
preserved the problem to try and analyze it.


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBCgAGBQJTmlDyAAoJEI5FoCIzSKrwN0IH/R6ny2h1XLv3Rj2Aed6C8xzU
+XsiNo6RYG9++Jk3vr705CVnmWx+4lwhr7E6jW9ValdDKEYjamypeQUyqrWSokiH
IajIMc4BNowGW7Eg1uRV8KQfb1P+QtmxBMwumOnOYk9zPfA9JcmlPVT6g5LrKy9N
6TMIlCMY1v0WPKZhseYYWHayP8PHF8KqSVuGIAZoPKPQdOQjNAOLW+PPbxiQKCcB
PrspPx+hNk4NUUm8156BdJO+xEGQpSYIY384yo0dEfvh3QHB4z/wTv84/9APxEQO
rV+67fW1psrFX+GkUolL9WFCBOeZRk3nykYJRravSOuOJKBa9laGAmiJT/EjWYI=
=DONL
-----END PGP SIGNATURE-----

Re: git rebase --skip stuck in a loop

[Jeff King]

On Thu, Jun 12, 2014 at 09:16:34PM -0400, Phillip Susi wrote:

> > I'm assuming this works in the general case, because we have tests
> > for it.  Do you perhaps have a repository and a set of reproduction
> > steps we could use to try to reproduce this?
> > 
> > If you can get that information, I'm interested in trying to fix
> > it.
> 
> I don't have a set of reproduction steps, but I do still have the repo
> that is in this stuck state if there is any data you might want me to
> pull out of it ( .git/rebase-apply? ).

Thanks for saving the stuck state.

If it's possible to share the whole repo, it might be worth seeing (then
we can all just run "git rebase --continue" ourselves). If it's too big
or is confidential, just tarring up .git/rebase-apply and making it
available is probably a good first step.

-Peff

Re: git rebase --skip stuck in a loop

[Phillip Susi]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 6/13/2014 3:34 AM, Jeff King wrote:
> Thanks for saving the stuck state.
> 
> If it's possible to share the whole repo, it might be worth seeing
> (then we can all just run "git rebase --continue" ourselves). If
> it's too big or is confidential, just tarring up .git/rebase-apply
> and making it available is probably a good first step.

It's the debian parted repo, so nothing confidential.  Here it is:

https://drive.google.com/file/d/0ByOQJBpP4bDXXy13YlN0aE5Fcnc/edit?usp=sharing


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.17 (MingW32)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJTmweVAAoJEI5FoCIzSKrwI84H/2Di1d5MeQpcg/I02nF7sd/f
gICGclFE8MPuTnKYfpYf4/SbSnB9+Znp+9OhQA0frYIblWHWEnUzwhINrDHvqApK
hCsuNGb5iHgy8ohZVGqH8B4ew8x4Ru3pwy9VYV2Wc1Z33curDIohx6qi4LCIrlaB
StzWgq3h8hV+3QB5zcD9MfAfdKkz1u5bIMlT9VYofYwctYERVk2DyaAj8eKBGyWo
6dWYuZDoQfO3Hnd/uGqWbdHrPiDoSQqYPbHNWNOnX24w+IKDv2Xe9bcHro+9A+Jg
PF+QXo4IS1E7CwevExe7gGKK3KdrD9SMixW7hHAbVVnpDIsvw485EL0xsZfmFJk=
=ncor
-----END PGP SIGNATURE-----

Re: git rebase --skip stuck in a loop

[brian m. carlson]

[-- Attachment #1: Type: text/plain, Size: 1172 bytes --]

On Fri, Jun 13, 2014 at 10:15:49AM -0400, Phillip Susi wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On 6/13/2014 3:34 AM, Jeff King wrote:
> > Thanks for saving the stuck state.
> > 
> > If it's possible to share the whole repo, it might be worth seeing
> > (then we can all just run "git rebase --continue" ourselves). If
> > it's too big or is confidential, just tarring up .git/rebase-apply
> > and making it available is probably a good first step.
> 
> It's the debian parted repo, so nothing confidential.  Here it is:
> 
> https://drive.google.com/file/d/0ByOQJBpP4bDXXy13YlN0aE5Fcnc/edit?usp=sharing

Okay.  This is actually a fairly straightforward issue.  In order to
reproduce, you have to use git rebase --merge and two patches in
sequence have to cause conflicts.  In that case, git rebase --skip will
fail to ever update the msgnum value, and it will get caught in a loop.

I should have a patch shortly.

-- 
brian m. carlson / brian with sandals: Houston, Texas, US
+1 832 623 2791 | http://www.crustytoothpaste.net/~bmc | My opinion only
OpenPGP: RSA v4 4096b: 88AC E9B2 9196 305B A994 7552 F1BA 225C 0223 B187

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

[PATCH] rebase--merge: fix --skip with two conflicts in a row

[brian m. carlson]

If git rebase --merge encountered a conflict, --skip would not work if the
next commit also conflicted.  The msgnum file would never be updated with
the new patch number, so no patch would actually be skipped, resulting in an
inescapable loop.

Update the msgnum file's value as the first thing in call_merge.  This also
avoids an "Already applied" message when skipping a commit.  There is no
visible change for the other contexts in which call_merge is invoked, as the
msgnum file's value remains unchanged in those situations.

Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
---
 git-rebase--merge.sh    |  5 +++--
 t/t3402-rebase-merge.sh | 15 +++++++++++++++
 2 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/git-rebase--merge.sh b/git-rebase--merge.sh
index 6d77b3c..d3fb67d 100644
--- a/git-rebase--merge.sh
+++ b/git-rebase--merge.sh
@@ -53,11 +53,12 @@ continue_merge () {
 }
 
 call_merge () {
-	cmt="$(cat "$state_dir/cmt.$1")"
+	msgnum="$1"
+	echo "$msgnum" >"$state_dir/msgnum"
+	cmt="$(cat "$state_dir/cmt.$msgnum")"
 	echo "$cmt" > "$state_dir/current"
 	hd=$(git rev-parse --verify HEAD)
 	cmt_name=$(git symbolic-ref HEAD 2> /dev/null || echo HEAD)
-	msgnum=$(cat "$state_dir/msgnum")
 	eval GITHEAD_$cmt='"${cmt_name##refs/heads/}~$(($end - $msgnum))"'
 	eval GITHEAD_$hd='$onto_name'
 	export GITHEAD_$cmt GITHEAD_$hd
diff --git a/t/t3402-rebase-merge.sh b/t/t3402-rebase-merge.sh
index be8c1d5..5a27ec9 100755
--- a/t/t3402-rebase-merge.sh
+++ b/t/t3402-rebase-merge.sh
@@ -33,6 +33,7 @@ test_expect_success setup '
 	tr "[a-z]" "[A-Z]" <original >newfile &&
 	git add newfile &&
 	git commit -a -m"side edits further." &&
+	git branch second-side &&
 
 	tr "[a-m]" "[A-M]" <original >newfile &&
 	rm -f original &&
@@ -41,6 +42,7 @@ test_expect_success setup '
 	git branch test-rebase side &&
 	git branch test-rebase-pick side &&
 	git branch test-reference-pick side &&
+	git branch test-conflicts side &&
 	git checkout -b test-merge side
 '
 
@@ -138,4 +140,17 @@ test_expect_success 'rebase -s funny -Xopt' '
 	test -f funny.was.run
 '
 
+test_expect_success 'rebase --skip works with two conflicts in a row' '
+	git checkout second-side  &&
+	tr "[A-Z]" "[a-z]" <newfile >tmp &&
+	mv tmp newfile &&
+	git commit -a -m"edit conflicting with side" &&
+	tr "[d-f]" "[D-F]" <newfile >tmp &&
+	mv tmp newfile &&
+	git commit -a -m"another edit conflicting with side" &&
+	test_must_fail git rebase --merge test-conflicts &&
+	test_must_fail git rebase --skip &&
+	git rebase --skip
+'
+
 test_done
-- 
2.0.0

Re: [PATCH] rebase--merge: fix --skip with two conflicts in a row

[Junio C Hamano]

"brian m. carlson" <sandals@crustytoothpaste.net> writes:

> If git rebase --merge encountered a conflict, --skip would not work if the
> next commit also conflicted.  The msgnum file would never be updated with
> the new patch number, so no patch would actually be skipped, resulting in an
> inescapable loop.
>
> Update the msgnum file's value as the first thing in call_merge.  This also
> avoids an "Already applied" message when skipping a commit.  There is no
> visible change for the other contexts in which call_merge is invoked, as the
> msgnum file's value remains unchanged in those situations.
>
> Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
> ---

Sounds good to me.  Thanks.

>  git-rebase--merge.sh    |  5 +++--
>  t/t3402-rebase-merge.sh | 15 +++++++++++++++
>  2 files changed, 18 insertions(+), 2 deletions(-)
>
> diff --git a/git-rebase--merge.sh b/git-rebase--merge.sh
> index 6d77b3c..d3fb67d 100644
> --- a/git-rebase--merge.sh
> +++ b/git-rebase--merge.sh
> @@ -53,11 +53,12 @@ continue_merge () {
>  }
>  
>  call_merge () {
> -	cmt="$(cat "$state_dir/cmt.$1")"
> +	msgnum="$1"
> +	echo "$msgnum" >"$state_dir/msgnum"
> +	cmt="$(cat "$state_dir/cmt.$msgnum")"
>  	echo "$cmt" > "$state_dir/current"
>  	hd=$(git rev-parse --verify HEAD)
>  	cmt_name=$(git symbolic-ref HEAD 2> /dev/null || echo HEAD)
> -	msgnum=$(cat "$state_dir/msgnum")
>  	eval GITHEAD_$cmt='"${cmt_name##refs/heads/}~$(($end - $msgnum))"'
>  	eval GITHEAD_$hd='$onto_name'
>  	export GITHEAD_$cmt GITHEAD_$hd
> diff --git a/t/t3402-rebase-merge.sh b/t/t3402-rebase-merge.sh
> index be8c1d5..5a27ec9 100755
> --- a/t/t3402-rebase-merge.sh
> +++ b/t/t3402-rebase-merge.sh
> @@ -33,6 +33,7 @@ test_expect_success setup '
>  	tr "[a-z]" "[A-Z]" <original >newfile &&
>  	git add newfile &&
>  	git commit -a -m"side edits further." &&
> +	git branch second-side &&
>  
>  	tr "[a-m]" "[A-M]" <original >newfile &&
>  	rm -f original &&
> @@ -41,6 +42,7 @@ test_expect_success setup '
>  	git branch test-rebase side &&
>  	git branch test-rebase-pick side &&
>  	git branch test-reference-pick side &&
> +	git branch test-conflicts side &&
>  	git checkout -b test-merge side
>  '
>  
> @@ -138,4 +140,17 @@ test_expect_success 'rebase -s funny -Xopt' '
>  	test -f funny.was.run
>  '
>  
> +test_expect_success 'rebase --skip works with two conflicts in a row' '
> +	git checkout second-side  &&
> +	tr "[A-Z]" "[a-z]" <newfile >tmp &&
> +	mv tmp newfile &&
> +	git commit -a -m"edit conflicting with side" &&
> +	tr "[d-f]" "[D-F]" <newfile >tmp &&
> +	mv tmp newfile &&
> +	git commit -a -m"another edit conflicting with side" &&
> +	test_must_fail git rebase --merge test-conflicts &&
> +	test_must_fail git rebase --skip &&
> +	git rebase --skip
> +'
> +
>  test_done