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

Re: [PATCH] [PATCH] Fix upload_pack_v2 response ordering for shallow fetch when server has uploadpack.allowRefInWant=true

From
Junio C Hamano <gitster@pobox.com>
Date
Sep 15, 2026, 20:19 UTC
Message-ID
<xmqq1pauz2ru.fsf@gitster.g>
In-Reply-To
<xmqqa4piz3pn.fsf@gitster.g>
Junio C Hamano <gitster@pobox.com> writes:
Show 14 quoted lines
>> diff --git a/upload-pack.c b/upload-pack.c
>> index a52856d869..a70d237ad3 100644
>> --- a/upload-pack.c
>> +++ b/upload-pack.c
>> @@ -1812,8 +1812,8 @@ int upload_pack_v2(struct repository *r, struct packet_reader *request)
>>  				state = UPLOAD_DONE;
>>  			break;
>>  		case UPLOAD_SEND_PACK:
>> -			send_wanted_ref_info(&data);
>>  			send_shallow_info(&data);
>> +			send_wanted_ref_info(&data);
>>  
>>  			if (data.uri_protocols.nr) {
>>  				create_pack_file(&data, &data.uri_protocols);

I am merely guessing what your reasoning is, but is this meant to match this part of the code on the other side of the connection?

		case FETCH_GET_PACK:
			trace2_region_leave("fetch-pack",
					    "negotiation_v2",
					    the_repository);
			trace2_data_intmax("negotiation_v2", the_repository,
					   "total_rounds", negotiation_round);
			/* Check for shallow-info section */
			if (process_section_header(&reader, "shallow-info", 1))
				receive_shallow_info(args, &reader, shallows, si);
			if (process_section_header(&reader, "wanted-refs", 1))
				receive_wanted_refs(&reader, sought, nr_sought);

These process_section_header() calls are made with the peek bit set, signaling that it is OK if the packet we are about to receive is not the one that is being checked, so what may happen is

 - upload-pack gives wanted-ref info and then shallow-info.
 - fetch-pack sees wanted-refs, notices that it is not shallow-info,
   ignores it, and the next process_section_header() call does
   notice it is wanted-refs and processes it.

But then, who consumes the shallow-info? Does fetch-pack notices shallow-info that it did not expect to see and crashes? If so, that is a very noteworthy thing to say in the proposed log message. If it does not crash and goes on but without utilizing what was carried in the shallow-info packet, the resulting behaviour of fetch-pack would be different from what we would expect, and if that is the case, that difference is a noteworthy thing to decribe in the proposed log message.

Thanks.
Previous: Junio C HamanoNext: Royce Gerard Remer
Message 3 of 7 in “Fix upload_pack_v2 response ordering for shallow fetch when server has uploadpack.allowRefInWant=true”
  1. [PATCH] Fix upload_pack_v2 response ordering for shallow fetch when server has uploadpack.allowRefInWant=trueRoyce Remer, Sep 15, 2026
  2. Junio C HamanoSep 15, 2026
  3. Junio C HamanoSep 15, 2026
  4. Royce Gerard RemerSep 15, 2026
  5. Junio C HamanoSep 16, 2026
  6. upload-pack: swap wanted-ref/shallow-info responsesRoyce Remer, Sep 16, 2026
  7. Junio C HamanoSep 17, 2026

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.