git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH] revision: mark blobs needed for resolve-undo as reachable

From
Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Date
Jun 14, 2022, 00:24 UTC
Message-ID
<220614.86czfcytlz.gmgdl@evledraar.gmail.com>
In-Reply-To
<7cd41846-e6ef-7a24-0426-6031a529360f@github.com>
On Mon, Jun 13 2022, Derrick Stolee wrote:
Show 17 quoted lines
> On 6/9/2022 7:44 PM, Junio C Hamano wrote:
>
>> +	struct string_list *resolve_undo = istate->resolve_undo;
>> +
>> +	if (!resolve_undo)
>> +		return 0;
>> +
>> +	for_each_string_list_item(item, resolve_undo) {
>
> I see this is necessary since for_each_string_list_item() does
> not handle NULL lists. After attempting to allow it to handle
> NULL lists, I see that the compiler complains about the cases
> where it would _never_ be NULL, so that change appears to be
> impossible.
>  
> The patch looks good. I liked the comments for the three phases
> of the test.

I think it's probably good to keep for_each_string_list_item() implemented the way it is, given that all existing callers of it feed non-NULL lists to it.

But why is it impossible to make it handle NULL lists? This works for
me, and passes the tests:
	
	diff --git a/builtin/fsck.c b/builtin/fsck.c
	index 4b17ccc3f44..4160bb50ad0 100644
	--- a/builtin/fsck.c
	+++ b/builtin/fsck.c
	@@ -763,9 +763,6 @@ static int fsck_resolve_undo(struct index_state *istate)
	 	struct string_list_item *item;
	 	struct string_list *resolve_undo = istate->resolve_undo;
	 
	-	if (!resolve_undo)
	-		return 0;
	-
	 	for_each_string_list_item(item, resolve_undo) {
	 		const char *path = item->string;
	 		struct resolve_undo_info *ru = item->util;
	diff --git a/string-list.h b/string-list.h
	index d5a744e1438..aa15aea8c2b 100644
	--- a/string-list.h
	+++ b/string-list.h
	@@ -143,7 +143,7 @@ int for_each_string_list(struct string_list *list,
	 
	 /** Iterate over each item, as a macro. */
	 #define for_each_string_list_item(item,list)            \
	-	for (item = (list)->items;                      \
	+	for (item = (((list) && (list)->items) ? ((list)->items) : NULL); \
	 	     item && item < (list)->items + (list)->nr; \
	 	     ++item)
	 

Perhaps you tried to convert it to a do/while macro with an "if", which won't work as we need the "for" to be used for the subsequent {} (or a single statement)..

But under the hood this is all just syntax sugar for "goto", so if we can avoid dereferencing "list" we're good to go...

Previous: Junio C HamanoNext: Derrick Stolee
Message 4 of 14 in “revision: mark blobs needed for resolve-undo as reachable”
  1. revision: mark blobs needed for resolve-undo as reachableJunio C Hamano, Jun 9, 2022
  2. Derrick StoleeJun 13, 2022
  3. Junio C HamanoJun 13, 2022
  4. Ævar Arnfjörð BjarmasonJun 14, 2022
  5. Derrick StoleeJun 14, 2022
  6. Taylor BlauJun 15, 2022
  7. Jeff KingJun 15, 2022
  8. Taylor BlauJun 15, 2022
  9. Junio C HamanoJun 15, 2022
  10. Ævar Arnfjörð BjarmasonJun 16, 2022
  11. Taylor BlauJun 14, 2022
  12. fsck segfault (was: Re: [PATCH] revision: mark blobs needed for resolve-undo as reachable)SZEDER Gábor, Jul 11, 2022
  13. Junio C HamanoJul 11, 2022
  14. 2/1 fsck: do not dereference NULL while checking resolve-undo dataJunio C Hamano, Jul 11, 2022

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.