Skip to contentSkip to conversation

refs: add missing remove_on_disk implementation for debug backend

PATCH In review RuanXinyu via GitGitGadget 4 participants 4 messages Oct 24, 2025 – Oct 27, 2025

lore
Share & tools
lore

RuanXinyu via GitGitGadget

author
#0 lore raw

PATCH 1/1 refs: add missing remove_on_disk implementation for debug backend

From: RuanXinyu <[email protected]>

The debug ref backend (refs_be_debug) was missing the remove_on_disk function pointer, which caused a segmentation fault when running 'GIT_TRACE_REFS=1 git refs migrate --ref-format=reftable' commands.

Signed-off-by Xinyu Ruan <[email protected]>

---
refs: add missing remove_on_disk implementation for debug backend

Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-git-2082%2FRuanXinyu%2Ffix-refs-trace-migrate-error-v1
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-git-2082/RuanXinyu/fix-refs-trace-migrate-error-v1
Pull-Request: https://github.com/git/git/pull/2082

 refs/debug.c | 9 +++++++++
 1 file changed, 9 insertions(+)
refs/debug.c +9 -0
index da300efaf3..dd49080836 100644
@@ -48,6 +48,14 @@ static int debug_create_on_disk(struct ref_store *refs, int flags, struct strbuf
 	return res;
 }
 
+static int debug_remove_on_disk(struct ref_store *refs, struct strbuf *err)
+{
+	struct debug_ref_store *drefs = (struct debug_ref_store *)refs;
+	int res = drefs->refs->be->remove_on_disk(drefs->refs, err);
+	trace_printf_key(&trace_refs, "remove_on_disk: %d\n", res);
+	return res;
+}
+
 static int debug_transaction_prepare(struct ref_store *refs,
 				     struct ref_transaction *transaction,
 				     struct strbuf *err)

@@ -432,6 +440,7 @@ struct ref_storage_be refs_be_debug = {
 	.init = NULL,
 	.release = debug_release,
 	.create_on_disk = debug_create_on_disk,
+	.remove_on_disk = debug_remove_on_disk,
 
 	/*
 	 * None of these should be NULL. If the "files" backend (in

base-commit: 81f86aacc4eb74cdb9c2c8082d36d2070c666045

Signature · 1 line
gitgitgadget
#1 lore raw
Quoted 5 lines · RuanXinyu via GitGitGadget

From: RuanXinyu <[email protected]>

The debug ref backend (refs_be_debug) was missing the remove_on_disk function pointer, which caused a segmentation fault when running 'GIT_TRACE_REFS=1 git refs migrate --ref-format=reftable' commands.

Heh, funny, just as I said nobody uses this infra you show up :) Good way to prove me wrong, thanks!

Quoted 1 line
Signed-off-by Xinyu Ruan <[email protected]>

Tiny nit: typically, the author and DCO should match. But the autor is "RuanXinyu" whereas the DCO says "Xinyu Ruan". I don't really think that this is something that warrants a new version, but I wanted to point this out anyway so that you can fix this going forward.

Quoted 20 lines
refs/debug.c +8 -0
index da300efaf3..dd49080836 100644
@@ -48,6 +48,14 @@ static int debug_create_on_disk(struct ref_store *refs, int flags, struct strbuf
 	return res;
 }
 
+static int debug_remove_on_disk(struct ref_store *refs, struct strbuf *err)
+{
+	struct debug_ref_store *drefs = (struct debug_ref_store *)refs;
+	int res = drefs->refs->be->remove_on_disk(drefs->refs, err);
+	trace_printf_key(&trace_refs, "remove_on_disk: %d\n", res);
+	return res;
+}
+
 static int debug_transaction_prepare(struct ref_store *refs,
 				     struct ref_transaction *transaction,
 				     struct strbuf *err)

refs/debug.c struct ref_storage_be refs_be_debug = { · line 446

Quoted 7 lines
refs/debug.c +1 -0
@@ -432,6 +440,7 @@ struct ref_storage_be refs_be_debug = {
 	.init = NULL,
 	.release = debug_release,
 	.create_on_disk = debug_create_on_disk,
+	.remove_on_disk = debug_remove_on_disk,
 
 	/*
 	 * None of these should be NULL. If the "files" backend (in

Yup, the implementation looks obviously correct to me. Thanks for fixing this bug!

Patrick

#2 lore raw
Quoted 6 lines · Patrick Steinhardt
Quoted 1 line
Signed-off-by Xinyu Ruan <[email protected]>

Tiny nit: typically, the author and DCO should match. But the autor is "RuanXinyu" whereas the DCO says "Xinyu Ruan". I don't really think that this is something that warrants a new version, but I wanted to point this out anyway so that you can fix this going forward.

It may not warrant a new version in the sense that I could tweak while queuing, but I need to be told which between the two is the name to be used before doing so.

I can make a guess and use the latter but it would cause me yet another piece of extra work if I guessed incorrectly, so...

#3 lore raw
Quoted 10 lines · Junio C Hamano
Quoted 6 lines · Patrick Steinhardt
Quoted 1 line
Signed-off-by Xinyu Ruan <[email protected]>

Tiny nit: typically, the author and DCO should match. But the autor is "RuanXinyu" whereas the DCO says "Xinyu Ruan". I don't really think that this is something that warrants a new version, but I wanted to point this out anyway so that you can fix this going forward.

Quoted 6 lines

It may not warrant a new version in the sense that I could tweak while queuing, but I need to be told which between the two is the name to be used before doing so.

I can make a guess and use the latter but it would cause me yet another piece of extra work if I guessed incorrectly, so...

Please use "Xinyu Ruan" for both author and Signed-off-by. Sorry for the confusion, and thanks for fixing it while queuing.