# git stash apply segfaulting when called in subdir 7 messages from 2007-07-25 to 2007-07-26. Participants: Uwe Kleine-König, Junio C Hamano, Steven Walter, Johannes Schindelin, Jeff King. Thread: https://gitlist.dev/t/9224 ## Uwe Kleine-König, 2007-07-25 21:23 Subject: git stash apply segfaulting when called in subdir Message-ID: <20070725212311.GA13325@cassiopeia> URL: https://gitlist.dev/e/20070725212311.GA13325%40cassiopeia ``` Hello, I did not do much testing and didn't look into the issue yet: zeisberg@cassiopeia:/tmp$ mkdir repo; cd repo; git init Initialized empty Git repository in .git/ zeisberg@cassiopeia:/tmp/repo$ mkdir dir; echo one > file; echo two > dir/file zeisberg@cassiopeia:/tmp/repo$ git add file dir/file zeisberg@cassiopeia:/tmp/repo$ git commit -m tralala Created initial commit 265b7d7: tralala 2 files changed, 2 insertions(+), 0 deletions(-) create mode 100644 dir/file create mode 100644 file zeisberg@cassiopeia:/tmp/repo$ echo three >> file zeisberg@cassiopeia:/tmp/repo$ git stash Saved "WIP on master: 265b7d7... tralala" HEAD is now at 265b7d7... tralala zeisberg@cassiopeia:/tmp/repo$ cd dir; git stash apply error: missing object referenced by '696146c2a44d7fc4d5ae4a71589c4c0d84f59789' /home/zeisberg/usr/bin/git-stash: line 111: 13618 Segmentation fault git-merge-recursive $b_tree -- $c_tree $w_tree Best regards Uwe -- Uwe Kleine-König http://www.google.com/search?q=1+newton+in+kg*m+%2F+s%5E2 ``` ## Junio C Hamano, 2007-07-25 22:28 Subject: Re: git stash apply segfaulting when called in subdir Message-ID: <7v1wewdsoz.fsf@assigned-by-dhcp.cox.net> URL: https://gitlist.dev/e/7v1wewdsoz.fsf%40assigned-by-dhcp.cox.net In-Reply-To: <20070725212311.GA13325@cassiopeia> ``` Uwe Kleine-König writes: > zeisberg@cassiopeia:/tmp$ mkdir repo; cd repo; git init > Initialized empty Git repository in .git/ > > zeisberg@cassiopeia:/tmp/repo$ mkdir dir; echo one > file; echo two > dir/file > > zeisberg@cassiopeia:/tmp/repo$ git add file dir/file > > zeisberg@cassiopeia:/tmp/repo$ git commit -m tralala > Created initial commit 265b7d7: tralala > 2 files changed, 2 insertions(+), 0 deletions(-) > create mode 100644 dir/file > create mode 100644 file > > zeisberg@cassiopeia:/tmp/repo$ echo three >> file > > zeisberg@cassiopeia:/tmp/repo$ git stash > Saved "WIP on master: 265b7d7... tralala" > HEAD is now at 265b7d7... tralala > > zeisberg@cassiopeia:/tmp/repo$ cd dir; git stash apply > error: missing object referenced by '696146c2a44d7fc4d5ae4a71589c4c0d84f59789' > /home/zeisberg/usr/bin/git-stash: line 111: 13618 Segmentation fault git-merge-recursive $b_tree -- $c_tree $w_tree This probably is a merge-recursive bug, but in the meantime, I think this should fix it. --- git-stash.sh | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/git-stash.sh b/git-stash.sh index de13dd1..d9cd42d 100755 --- a/git-stash.sh +++ b/git-stash.sh @@ -6,6 +6,7 @@ USAGE='[ | list | show | apply | clear]' SUBDIRECTORY_OK=Yes . git-sh-setup require_work_tree +cd_to_toplevel TMP="$GIT_DIR/.git-stash.$$" trap 'rm -f "$TMP-*"' 0 ``` ## Steven Walter, 2007-07-25 22:32 Subject: Re: git stash apply segfaulting when called in subdir Message-ID: <20070725223244.GB21274@dervierte> URL: https://gitlist.dev/e/20070725223244.GB21274%40dervierte In-Reply-To: <20070725212311.GA13325@cassiopeia> ``` On Wed, Jul 25, 2007 at 11:23:11PM +0200, Uwe Kleine-König wrote: > zeisberg@cassiopeia:/tmp/repo$ cd dir; git stash apply > error: missing object referenced by '696146c2a44d7fc4d5ae4a71589c4c0d84f59789' > /home/zeisberg/usr/bin/git-stash: line 111: 13618 Segmentation fault git-merge-recursive $b_tree -- $c_tree $w_tree I've been seeing this, as well. -- -Steven Walter "A human being should be able to change a diaper, plan an invasion, butcher a hog, conn a ship, design a building, write a sonnet, balance accounts, build a wall, set a bone, comfort the dying, take orders, give orders, cooperate, act alone, solve equations, analyze a new problem, pitch manure, program a computer, cook a tasty meal, fight efficiently, die gallantly. Specialization is for insects." -Robert Heinlein ``` ## Johannes Schindelin, 2007-07-25 23:01 Subject: Re: git stash apply segfaulting when called in subdir Message-ID: URL: https://gitlist.dev/e/Pine.LNX.4.64.0707260001091.14781%40racer.site In-Reply-To: <7v1wewdsoz.fsf@assigned-by-dhcp.cox.net> ``` Hi, On Wed, 25 Jul 2007, Junio C Hamano wrote: > Uwe Kleine-König writes: > > > zeisberg@cassiopeia:/tmp$ mkdir repo; cd repo; git init > > Initialized empty Git repository in .git/ > > > > zeisberg@cassiopeia:/tmp/repo$ mkdir dir; echo one > file; echo two > dir/file > > > > zeisberg@cassiopeia:/tmp/repo$ git add file dir/file > > > > zeisberg@cassiopeia:/tmp/repo$ git commit -m tralala > > Created initial commit 265b7d7: tralala > > 2 files changed, 2 insertions(+), 0 deletions(-) > > create mode 100644 dir/file > > create mode 100644 file > > > > zeisberg@cassiopeia:/tmp/repo$ echo three >> file > > > > zeisberg@cassiopeia:/tmp/repo$ git stash > > Saved "WIP on master: 265b7d7... tralala" > > HEAD is now at 265b7d7... tralala > > > > zeisberg@cassiopeia:/tmp/repo$ cd dir; git stash apply > > error: missing object referenced by '696146c2a44d7fc4d5ae4a71589c4c0d84f59789' > > /home/zeisberg/usr/bin/git-stash: line 111: 13618 Segmentation fault git-merge-recursive $b_tree -- $c_tree $w_tree > > This probably is a merge-recursive bug, but in the meantime, > I think this should fix it. > > --- > > git-stash.sh | 1 + > 1 files changed, 1 insertions(+), 0 deletions(-) > > diff --git a/git-stash.sh b/git-stash.sh > index de13dd1..d9cd42d 100755 > --- a/git-stash.sh > +++ b/git-stash.sh > @@ -6,6 +6,7 @@ USAGE='[ | list | show | apply | clear]' > SUBDIRECTORY_OK=Yes > . git-sh-setup > require_work_tree > +cd_to_toplevel > > TMP="$GIT_DIR/.git-stash.$$" > trap 'rm -f "$TMP-*"' 0 It does... And here is a test case: -- snipsnap -- [PATCH] Test "stash apply" in a subdirectory Signed-off-by: Johannes Schindelin --- t/t3903-stash.sh | 14 ++++++++++++++ 1 files changed, 14 insertions(+), 0 deletions(-) diff --git a/t/t3903-stash.sh b/t/t3903-stash.sh index 392ac1c..9e363fd 100755 --- a/t/t3903-stash.sh +++ b/t/t3903-stash.sh @@ -66,4 +66,18 @@ test_expect_success 'apply stashed changes (including index)' ' test 1 = $(git show HEAD:file) ' +test_expect_success 'apply in subdir' ' + git --git-dir=new init && + cd new && + mkdir dir && + echo one > file && + echo two > dir/file && + git add file dir/file && + git commit -m first && + echo three >> file && + git stash && + cd dir && + git stash apply +' + test_done ``` ## Jeff King, 2007-07-26 05:07 Subject: Re: git stash apply segfaulting when called in subdir Message-ID: <20070726050726.GC32617@coredump.intra.peff.net> URL: https://gitlist.dev/e/20070726050726.GC32617%40coredump.intra.peff.net In-Reply-To: <7v1wewdsoz.fsf@assigned-by-dhcp.cox.net> ``` On Wed, Jul 25, 2007 at 03:28:12PM -0700, Junio C Hamano wrote: > > zeisberg@cassiopeia:/tmp/repo$ cd dir; git stash apply > > error: missing object referenced by '696146c2a44d7fc4d5ae4a71589c4c0d84f59789' > > /home/zeisberg/usr/bin/git-stash: line 111: 13618 Segmentation fault git-merge-recursive $b_tree -- $c_tree $w_tree > > This probably is a merge-recursive bug, but in the meantime, > I think this should fix it. The merge-recursive bug is simply a failure to check for invalid input. Patch to at least print and die rather than segfault is below. Technically the error could be from a tag dereference, too, so maybe the error message should be more exact? -- >8 -- merge-recursive: don't segfault on missing input objects This should generally never happen, but it's nicer to report an error than to segfault. Signed-off-by: Jeff King --- merge-recursive.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-) diff --git a/merge-recursive.c b/merge-recursive.c index c8539ec..2764e61 100644 --- a/merge-recursive.c +++ b/merge-recursive.c @@ -1692,6 +1692,8 @@ static struct commit *get_ref(const char *ref) if (get_sha1(ref, sha1)) die("Could not resolve ref '%s'", ref); object = deref_tag(parse_object(sha1), ref, strlen(ref)); + if (!object) + die("Object does not exist: %s", ref); if (object->type == OBJ_TREE) return make_virtual_commit((struct tree*)object, better_branch_name(ref)); -- 1.5.3.rc3.818.gc1dc7-dirty ``` ## Johannes Schindelin, 2007-07-26 05:40 Subject: Re: git stash apply segfaulting when called in subdir Message-ID: URL: https://gitlist.dev/e/Pine.LNX.4.64.0707260635060.14781%40racer.site In-Reply-To: <20070726050726.GC32617@coredump.intra.peff.net> ``` Hi, On Thu, 26 Jul 2007, Jeff King wrote: > On Wed, Jul 25, 2007 at 03:28:12PM -0700, Junio C Hamano wrote: > > > > zeisberg@cassiopeia:/tmp/repo$ cd dir; git stash apply > > > error: missing object referenced by '696146c2a44d7fc4d5ae4a71589c4c0d84f59789' > > > /home/zeisberg/usr/bin/git-stash: line 111: 13618 Segmentation fault git-merge-recursive $b_tree -- $c_tree $w_tree > > > > This probably is a merge-recursive bug, but in the meantime, > > I think this should fix it. > > The merge-recursive bug is simply a failure to check for invalid input. No it is not. The objects are there. But the objects cannot be found, since merge-recursive cannot go to top level. Ciao, Dscho ``` ## Jeff King, 2007-07-26 05:48 Subject: Re: git stash apply segfaulting when called in subdir Message-ID: <20070726054800.GA21518@coredump.intra.peff.net> URL: https://gitlist.dev/e/20070726054800.GA21518%40coredump.intra.peff.net In-Reply-To: ``` On Thu, Jul 26, 2007 at 06:40:44AM +0100, Johannes Schindelin wrote: > > The merge-recursive bug is simply a failure to check for invalid input. > > No it is not. > > The objects are there. But the objects cannot be found, since > merge-recursive cannot go to top level. Fair enough, I should have looked a little further. However, I still wonder if it's better to report an error than to segfault. -Peff ```