In the Git pack protocol definition, an error packet may appear only in
a certain context. However, servers can face a runtime error (e.g. I/O
error) at an arbitrary timing. This patch changes the protocol to allow
an error packet to be sent instead of any packet.
Without this protocol spec change, when a server cannot process a
request, there's no way to tell that to a client. Since the server
cannot produce a valid response, it would be forced to cut a connection
without telling why. With this protocol spec change, the server can be
more gentle in this situation. An old client may see these error packets
as an unexpected packet, but this is not worse than having an unexpected
EOF.
Following this protocol spec change, the error packet handling code is
moved to pkt-line.c. Implementation wise, this implementation uses
pkt-line to communicate with a subprocess. Since this is not a part of
Git protocol, it's possible that a packet that is not supposed to be an
error packet is mistakenly parsed as an error packet. This error packet
handling is enabled only for the Git pack protocol parsing code
considering this.
Signed-off-by: Masaya Suzuki <masayasuzuki@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
By using and sharing a packet_reader while handling a Git pack protocol
request, the same reader option is used throughout the code. This makes
it easy to set a reader option to the request parsing code.
Signed-off-by: Masaya Suzuki <masayasuzuki@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Commit 024aa4696c (fetch-pack.c: use oidset to check existence of loose
object, 2018-03-14) added a cache to avoid calling stat() for a bunch of
loose objects we don't have.
Now that OBJECT_INFO_QUICK handles this caching itself, we can drop the
custom solution.
Note that this might perform slightly differently, as the original code
stopped calling readdir() when we saw more loose objects than there were
refs. So:
1. The old code might have spent work on readdir() to fill the cache,
but then decided there were too many loose objects, wasting that
effort.
2. The new code might spend a lot of time on readdir() if you have a
lot of loose objects, even though there are very few objects to
ask about.
In practice it probably won't matter either way; see the previous commit
for some discussion of the tradeoff.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Each section in a protocol v2 response is followed by either a DELIM
packet (indicating more sections to follow) or a FLUSH packet
(indicating none to follow). But when parsing the "acknowledgments"
section, do_fetch_pack_v2() is liberal in accepting both, but determines
whether to continue reading or not based solely on the contents of the
"acknowledgments" section, not on whether DELIM or FLUSH was read.
There is no issue with a protocol-compliant server, but can result in
confusing error messages when communicating with a server that
serves unexpected additional sections. Consider a server that sends
"new-section" after "acknowledgments":
- client writes request
- client reads the "acknowledgments" section which contains no "ready",
then DELIM
- since there was no "ready", client needs to continue negotiation, and
writes request
- client reads "new-section", and reports to the end user "expected
'acknowledgments', received 'new-section'"
For the person debugging the involved Git implementation(s), the error
message is confusing in that "new-section" was not received in response
to the latest request, but to the first one.
One solution is to always continue reading after DELIM, but in this
case, we can do better. We know from the protocol that "ready" means at
least the packfile section is coming (hence, DELIM) and that no "ready"
means that no sections are to follow (hence, FLUSH). So teach
process_acks() to enforce this.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
None of the current callers use the refname parameter we pass to their
callbacks. In theory somebody _could_ do so, but it's actually quite
weird if you think about it: it's a ref in somebody else's repository.
So the name has no meaning locally, and in fact there may be duplicates
if there are multiple alternates.
The users of this interface really only care about seeing some ref tips,
since that promises that the alternate has the full commit graph
reachable from there. So let's keep the information we pass back to the
bare minimum.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Acked-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When all refs to be fetched are exact OIDs, it is possible to perform a
fetch without requiring the remote to list refs if protocol v2 is used.
Teach Git to do this.
This currently has an effect only for lazy fetches done from partial
clones. The change necessary to likewise optimize "git fetch <remote>
<sha-1>" will be done in a subsequent patch.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
tip_oids_contain() lazily loads refs into an oidset at its first call.
It abuses the internal (sub)member .map.tablesize of that oidset to
check if it has done that already.
Determine if the oidset needs to be populated upfront and then do that
instead. This duplicates a loop, but simplifies the existing one by
separating concerns between the two.
Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Move the code to determine if a request is unmatched to its own little
helper. This allows us to reuse it in a subsequent patch.
Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
A partial clone with missing trees can be obtained using "git clone
--filter=tree:none <repo>". In such a repository, when a tree needs to
be lazily fetched, any tree or blob it directly or indirectly references
is fetched as well, regardless of whether the original command required
those objects, or if the local repository already had some of them.
This is because the fetch protocol, which the lazy fetch uses, does not
allow clients to request that only the wanted objects be sent, which
would be the ideal solution. This patch implements a partial solution:
specify the "blob:none" filter, somewhat reducing the fetch payload.
This change has no effect when lazily fetching blobs (due to how filters
work). And if lazily fetching a commit (such repositories are difficult
to construct and is not a use case we support very well, but it is
possible), referenced commits and trees are still fetched - only the
blobs are not fetched.
The necessary code change is done in fetch_pack() instead of somewhere
closer to where the "filter" instruction is written to the wire so that
only one part of the code needs to be changed in order for users of all
protocol versions to benefit from this optimization.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When fetch_pack() is invoked as part of another Git command (due to a
lazy fetch from a partial clone, for example), it uses object flags that
may already be used by the outer Git command.
The commit that introduced the lazy fetch feature (88e2f9ed8e
("introduce fetch-object: fetch one promisor object", 2017-12-05)) tried
to avoid this overlap, but it did not avoid it totally. It was
successful in avoiding writing COMPLETE, but did not avoid reading
COMPLETE, and did not avoid writing and reading ALTERNATE.
Ensure that no flags are written or read by fetch_pack() in the case
where it is used to perform a lazy fetch. To do this, it is sufficient
to avoid checking completeness of wanted refs (unnecessary in the case
of lazy fetches), and to avoid negotiation-related work (in the current
implementation, already, no negotiation is performed). After that was
done, the lack of overlap was verified by checking all direct and
indirect usages of COMPLETE and ALTERNATE - that they are read or
written only if no_dependents is false.
There are other possible solutions to this issue:
(1) Split fetch-pack.{c,h} into a flag-using part and a non-flag-using
part, and whenever no_dependents is set, only use the
non-flag-using part.
(2) Make fetch_pack() be able to be used with arbitrary repository
objects. fetch_pack() should then create its own repository object
based on the given repository object, with its own object
hashtable, so that the flags do not conflict.
(1) is possible but invasive - some functions would need to be split;
and such invasiveness would potentially be unnecessary if we ever were
to need (2) anyway. (2) would be useful if we were to support, say,
submodules that were partial clones themselves, but I don't know when or
if the Git project plans to support those.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This is the flip side of the previous two patches: checking
for a non-zero oidcmp() can be more strictly expressed as
inequality. Like those patches, we write "!= 0" in the
coccinelle transformation, which covers by isomorphism the
more common:
if (oidcmp(E1, E2))
As with the previous two patches, this patch can be achieved
almost entirely by running "make coccicheck"; the only
differences are manual line-wrap fixes to match the original
code.
There is one thing to note for anybody replicating this,
though: coccinelle 1.0.4 seems to miss the case in
builtin/tag.c, even though it's basically the same as all
the others. Running with 1.0.7 does catch this, so
presumably it's just a coccinelle bug that was fixed in the
interim.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When a user fetches:
- at least one up-to-date ref and at least one non-up-to-date ref,
- using HTTP with protocol v0 (or something else that uses the fetch
command of a remote helper)
some refs might not be updated after the fetch.
This bug was introduced in commit 989b8c4452 ("fetch-pack: put shallow
info in output parameter", 2018-06-28) which allowed transports to
report the refs that they have fetched in a new out-parameter
"fetched_refs". If they do so, transport_fetch_refs() makes this
information available to its caller.
Users of "fetched_refs" rely on the following 3 properties:
(1) it is the complete list of refs that was passed to
transport_fetch_refs(),
(2) it has shallow information (REF_STATUS_REJECT_SHALLOW set if
relevant), and
(3) it has updated OIDs if ref-in-want was used (introduced after
989b8c4452).
In an effort to satisfy (1), whenever transport_fetch_refs()
filters the refs sent to the transport, it re-adds the filtered refs to
whatever the transport supplies before returning it to the user.
However, the implementation in 989b8c4452 unconditionally re-adds the
filtered refs without checking if the transport refrained from reporting
anything in "fetched_refs" (which it is allowed to do), resulting in an
incomplete list, no longer satisfying (1).
An earlier effort to resolve this [1] solved the issue by readding the
filtered refs only if the transport did not refrain from reporting in
"fetched_refs", but after further discussion, it seems that the better
solution is to revert the API change that introduced "fetched_refs".
This API change was first suggested as part of a ref-in-want
implementation that allowed for ref patterns and, thus, there could be
drastic differences between the input refs and the refs actually fetched
[2]; we eventually decided to only allow exact ref names, but this API
change remained even though its necessity was decreased.
Therefore, revert this API change by reverting commit 989b8c4452, and
make receive_wanted_refs() update the OIDs in the sought array (like how
update_shallow() updates shallow information in the sought array)
instead. A test is also included to show that the user-visible bug
discussed at the beginning of this commit message no longer exists.
[1] https://public-inbox.org/git/20180801171806.GA122458@google.com/
[2] https://public-inbox.org/git/86a128c5fb710a41791e7183207c4d64889f9307.1485381677.git.jonathantanmy@google.com/
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Implement support for fetch.fsck.* corresponding with the existing
receive.fsck.*. This allows for pedantically cloning repositories with
specific issues without turning off fetch.fsckObjects.
One such repository is https://github.com/robbyrussell/oh-my-zsh.git
which before this change will emit this error when cloned with
fetch.fsckObjects:
error: object 2b7227859263b6aabcc28355b0b994995b7148b6: zeroPaddedFilemode: contains zero-padded file modes
fatal: Error in object
fatal: index-pack failed
Now with fetch.fsck.zeroPaddedFilemode=warn we'll warn about that
issue, but the clone will succeed:
warning: object 2b7227859263b6aabcc28355b0b994995b7148b6: zeroPaddedFilemode: contains zero-padded file modes
warning: object a18c4d13c2a5fa2d4ecd5346c50e119b999b807d: zeroPaddedFilemode: contains zero-padded file modes
warning: object 84df066176c8da3fd59b13731a86d90f4f1e5c9d: zeroPaddedFilemode: contains zero-padded file modes
The motivation for this is to be able to turn on fetch.fsckObjects
globally across a fleet of computers but still be able to manually
clone various legacy repositories by either white-listing specific
issues, or better yet whitelist specific objects.
The use of --git-dir=* instead of -C in the tests could be considered
somewhat archaic, but the tests I'm adding here are duplicating the
corresponding receive.* tests with as few changes as possible.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Introduce a new negotiation algorithm used during fetch that skips
commits in an effort to find common ancestors faster. The skips grow
similarly to the Fibonacci sequence as the commit walk proceeds further
away from the tips. The skips may cause unnecessary commits to be
included in the packfile, but the negotiation step typically ends more
quickly.
Usage of this algorithm is guarded behind the configuration flag
fetch.negotiationAlgorithm.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
During negotiation, fetch-pack eventually reports as "have" lines all
commits reachable from all refs. Allow the user to restrict the commits
sent in this way by providing a whitelist of tips; only the tips
themselves and their ancestors will be sent.
Both globs and single objects are supported.
This feature is only supported for protocols that support connect or
stateless-connect (such as HTTP with protocol v2).
This will speed up negotiation when the repository has multiple
relatively independent branches (for example, when a repository
interacts with multiple repositories, such as with linux-next [1] and
torvalds/linux [2]), and the user knows which local branch is likely to
have commits in common with the upstream branch they are fetching.
[1] https://kernel.googlesource.com/pub/scm/linux/kernel/git/next/linux-next/
[2] https://kernel.googlesource.com/pub/scm/linux/kernel/git/torvalds/linux/
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When fetching, connectivity is checked after the shallow file is
updated. There are 2 issues with this: (1) the connectivity check is
only performed up to ancestors of existing refs (which is not thorough
enough if we were deepening an existing ref in the first place), and (2)
there is no rollback of the shallow file if the connectivity check
fails.
To solve (1), update the connectivity check to check the ancestry chain
completely in the case of a deepening fetch by refraining from passing
"--not --all" when invoking rev-list in connected.c.
To solve (2), have fetch_pack() perform its own connectivity check
before updating the shallow file. To support existing use cases in which
"git fetch-pack" is used to download objects without much regard as to
the connectivity of the resulting objects with respect to the existing
repository, the connectivity check is only done if necessary (that is,
the fetch is not a clone, and the fetch involves shallow/deepen
functionality). "git fetch" still performs its own connectivity check,
preserving correctness but sometimes performing redundant work. This
redundancy is mitigated by the fact that fetch_pack() reports if it has
performed a connectivity check itself, and if the transport supports
connect or stateless-connect, it will bubble up that report so that "git
fetch" knows not to perform the connectivity check in such a case.
This was noticed when a user tried to deepen an existing repository by
fetching with --no-shallow from a server that did not send all necessary
objects - the connectivity check as run by "git fetch" succeeded, but a
subsequent "git fsck" failed.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a repository argument to allow the callers of deref_tag
to be more specific about which repository to act on. This is a small
mechanical change; it doesn't change the implementation to handle
repositories other than the_repository yet.
As with the previous commits, use a macro to catch callers passing a
repository other than the_repository at compile time.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a repository argument to allow callers of lookup_commit to be more
specific about which repository to handle. This is a small mechanical
change; it doesn't change the implementation to handle repositories
other than the_repository yet.
As with the previous commits, use a macro to catch callers passing a
repository other than the_repository at compile time.
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a repository argument to allow callers of lookup_object to be more
specific about which repository to handle. This is a small mechanical
change; it doesn't change the implementation to handle repositories
other than the_repository yet.
As with the previous commits, use a macro to catch callers passing a
repository other than the_repository at compile time.
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a repository argument to allow the callers of parse_object
to be more specific about which repository to act on. This is a small
mechanical change; it doesn't change the implementation to handle
repositories other than the_repository yet.
As with the previous commits, use a macro to catch callers passing a
repository other than the_repository at compile time.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Implement ref-in-want on the client side so that when a server supports
the "ref-in-want" feature, a client will send "want-ref" lines for each
reference the client wants to fetch. This feature allows clients to
tolerate inconsistencies that exist when a remote repository's refs
change during the course of negotiation.
This allows a client to request to request a particular ref without
specifying the OID of the ref. This means that instead of hitting an
error when a ref no longer points at the OID it did at the beginning of
negotiation, negotiation can continue and the value of that ref will be
sent at the termination of negotiation, just before a packfile is sent.
More information on the ref-in-want feature can be found in
Documentation/technical/protocol-v2.txt.
Signed-off-by: Brandon Williams <bmwill@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Expand the transport fetch method signature, by adding an output
parameter, to allow transports to return information about the refs they
have fetched. Then communicate shallow status information through this
mechanism instead of by modifying the input list of refs.
This does require clients to sometimes generate the ref map twice: once
from the list of refs provided by the remote (as is currently done) and
potentially once from the new list of refs that the fetch mechanism
provides.
Signed-off-by: Brandon Williams <bmwill@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Introduce the new files fetch-negotiator.{h,c}, which contains an API
behind which the details of negotiation are abstracted. Currently, only
one algorithm is available: the existing one.
This patch is written to be easily reviewed: static functions are
moved verbatim from fetch-pack.c to negotiator/default.c, and it can be
seen that the lines replaced by negotiator->X() calls are present in the
X() functions respectively.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When receiving 'ACK <object-id> continue' for a common commit, check if
the commit was already known to be common and mark it as such if not up
front. This should make future refactoring of how the information about
common commits is stored more straightforward.
No visible change intended.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Reduce the number of global variables by making the priority queue and
the count of non-common commits in it local, passing them as a struct to
various functions where necessary.
This also helps in the case that fetch_pack() is invoked twice in the
same process (when tag following is required when using a transport that
does not support tag following), in that different priority queues will
now be used in each invocation, instead of reusing the possibly
non-empty one.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In negotiation using protocol v2, fetch-pack sometimes does not make
full use of the information obtained in the ref advertisement:
specifically, that if the server advertises a commit that the client
also has, the client never needs to inform the server that it has the
commit's parents, since it can just tell the server that it has the
advertised commit and it knows that the server can and will infer the
rest.
This is because, in do_fetch_pack_v2(), rev_list_insert_ref_oid() is
invoked before mark_complete_and_common_ref(). This means that if we
have a commit that is both our ref and their ref, it would be enqueued
by rev_list_insert_ref_oid() as SEEN, and since it is thus already SEEN,
mark_complete_and_common_ref() would not enqueue it.
If mark_complete_and_common_ref() were invoked first, as it is in
do_fetch_pack() for protocol v0, then mark_complete_and_common_ref()
would enqueue it with COMMON_REF | SEEN. The addition of COMMON_REF
ensures that its parents are not sent as "have" lines.
Change the order in do_fetch_pack_v2() to be consistent with
do_fetch_pack(), and to avoid sending unnecessary "have" lines.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When "ACK %s ready" is received, find_common() clears rev_list in an
attempt to stop further "have" lines from being sent [1]. It is much
more readable to explicitly break from the loop instead.
So explicitly break from the loop, and make the clearing of the rev_list
happen unconditionally.
[1] The rationale is further described in the originating commit
f2cba9299b ("fetch-pack: Finish negotation if remote replies "ACK %s
ready"", 2011-03-14).
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If tag following is required when using a transport that does not
support tag following, fetch_pack() will be invoked twice in the same
process, necessitating a clearing of the object flags used by
fetch_pack() sometime during the second invocation. This is currently
done in find_common(), which means that the invocation of
mark_complete_and_common_ref() in do_fetch_pack() is useless.
(This cannot be reproduced with Git alone, because all transports that
come with Git support tag following.)
Therefore, move this clearing from find_common() to its
parent function do_fetch_pack(), right before it calls
mark_complete_and_common_ref().
This has been occurring since the commit that introduced the clearing of
marks, 420e9af498 ("Fix tag following", 2008-03-19).
The corresponding code for protocol v2 in do_fetch_pack_v2() does not
have this problem, as the clearing of flags is done before any marking
(whether by rev_list_insert_ref_oid() or
mark_complete_and_common_ref()).
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The function everything_local(), despite its name, also (1) marks
commits as COMPLETE and COMMON_REF and (2) invokes filter_refs() as
important side effects. Extract (1) into its own function
(mark_complete_and_common_ref()) and remove
(2).
The restoring of save_commit_buffer, which was introduced in a1c6d7c1a7
("fetch-pack: restore save_commit_buffer after use", 2017-12-08), is a
concern of the parse_object() call in mark_complete_and_common_ref(), so
it has been moved from the end of everything_local() to the end of
mark_complete_and_common_ref().
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When "fetch-pack --all" sees a tag-to-blob on the remote, it
tries to fetch both the tag itself ("refs/tags/foo") and the
peeled value that the remote advertises ("refs/tags/foo^{}").
Asking for the object pointed to by the latter can cause
upload-pack to complain with "not our ref", since it does
not mark the peeled objects with the OUR_REF (unless they
were at the tip of some other ref).
Arguably upload-pack _should_ be marking those peeled
objects. But it never has in the past, since clients would
generally just ask for the tag and expect to get the peeled
value along with it. And that's how "git fetch" works, as
well as older versions of "fetch-pack --all".
The problem was introduced by 5f0fc64513 (fetch-pack:
eliminate spurious error messages, 2012-09-09). Before then,
the matching logic was something like:
if (refname is ill-formed)
do nothing
else if (doing --all)
always consider it matched
else
look through list of sought refs for a match
That commit wanted to flip the order of the second two arms
of that conditional. But we ended up with:
if (refname is ill-formed)
do nothing
else
look through list of sought refs for a match
if (--all and no match so far)
always consider it matched
That means tha an ill-formed ref will trigger the --all
conditional block, even though we should just be ignoring
it. We can fix that by having a single "else" with all of
the well-formed logic, that checks the sought refs and
"--all" in the correct order.
Reported-by: Kirill Smelkov <kirr@nexedi.com>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Migrate all git_path_* functions that are defined in path.c to take a
repository argument. Unlike other patches in this series, do not use the
#define trick, as we rewrite the whole function, which is rather small.
This doesn't migrate all the functions, as other builtins have their own
local path functions defined using GIT_PATH_FUNC. So keep that macro
around to serve the other locations.
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a repository argument to allow callers of is_repository_shallow
to be more specific about which repository to handle. This is a small
mechanical change; it doesn't change the implementation to handle
repositories other than the_repository yet.
As with the previous commits, use a macro to catch callers passing a
repository other than the_repository at compile time.
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a repository argument to allow callers of register_shallow
to be more specific about which repository to handle. This is a small
mechanical change; it doesn't change the implementation to handle
repositories other than the_repository yet.
As with the previous commits, use a macro to catch callers passing a
repository other than the_repository at compile time.
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This should make these functions easier to find and cache.h less
overwhelming to read.
In particular, this moves:
- read_object_file
- oid_object_info
- write_object_file
As a result, most of the codebase needs to #include object-store.h.
In this patch the #include is only added to files that would fail to
compile otherwise. It would be better to #include wherever
identifiers from the header are used. That can happen later
when we have better tooling for it.
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The fetch-pack/upload-pack protocol v2 was developed independently of
the filter parameter (used in partial fetches), thus it did not include
support for it. Add support for the filter parameter.
Like in the legacy protocol, the server advertises and supports "filter"
only if uploadpack.allowfilter is configured.
Like in the legacy protocol, the client continues with a warning if
"--filter" is specified, but the server does not advertise it.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Teach fetch to optionally accept server options by specifying them on
the cmdline via '-o' or '--server-option'. These server options are
sent to the remote end when performing a fetch communicating using
protocol version 2.
If communicating using a protocol other than v2 the provided options are
ignored and not sent to the remote end.
Signed-off-by: Brandon Williams <bmwill@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This is more consistent with the project style. The majority of Git's
source files use dashes in preference to underscores in their file names.
Signed-off-by: Stefan Beller <sbeller@google.com>
See previous patch for explanation.
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Enable shallow clones and deepen requests using protocol version 2 if
the server 'fetch' command supports the 'shallow' feature.
Signed-off-by: Brandon Williams <bmwill@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When communicating with a v2 server, perform a fetch by requesting the
'fetch' command.
Signed-off-by: Brandon Williams <bmwill@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When doing a partial clone or fetch with transfer.fsckobjects=1, use the
--fsck-objects instead of the --strict flag when invoking index-pack so
that links are not checked, only objects. This is because incomplete
links are expected when doing a partial clone or fetch.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When fetching from a repository with large number of refs, because to
check existence of each refs in local repository to packed and loose
objects, 'git fetch' ends up doing a lot of lstat(2) to non-existing
loose form, which makes it slow.
Instead of making as many lstat(2) calls as the refs the remote side
advertised to see if these objects exist in the loose form, first
enumerate all the existing loose objects in hashmap beforehand and use
it to check existence of them if the number of refs is larger than the
number of loose objects.
With this patch, the number of lstat(2) calls in `git fetch` is reduced
from 411412 to 13794 for chromium repository, it has more than 480000
remote refs.
I took time stat of `git fetch` when fetch-pack happens for chromium
repository 3 times on linux with SSD.
* with this patch
8.105s
8.309s
7.640s
avg: 8.018s
* master
12.287s
11.175s
12.227s
avg: 11.896s
On my MacBook Air which has slower lstat(2).
* with this patch
14.501s
* master
1m16.027s
`git fetch` on slow disk will be improved largely.
Signed-off-by: Takuto Ikuta <tikuta@chromium.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The packet_read_line() function dies if it gets an
unexpected EOF. It only returns NULL if we get a flush
packet (or technically, a zero-length "0004" packet, but
nobody is supposed to send those, and they are
indistinguishable from a flush in this interface).
Let's correct error messages which claim an unexpected EOF;
it's really an unexpected flush packet.
While we're here, let's also check "!line" instead of
"!len" in the second case. The two events should always
coincide, but checking "!line" makes it more obvious that we
are not about to dereference NULL.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In fetch-pack, the global variable save_commit_buffer is set to 0, but
not restored to its original value after use.
In particular, if show_log() (in log-tree.c) is invoked after
fetch_pack() in the same process, show_log() will return before printing
out the commit message (because the invocation to
get_cached_commit_buffer() returns NULL, because the commit buffer was
not saved). I discovered this when attempting to run "git log -S" in a
partial clone, triggering the case where revision walking lazily loads
missing objects.
Therefore, restore save_commit_buffer to its original value after use.
An alternative to solve the problem I had is to replace
get_cached_commit_buffer() with get_commit_buffer(). That invocation was
introduced in commit a97934d ("use get_cached_commit_buffer where
appropriate", 2014-06-13) to replace "commit->buffer" introduced in
commit 3131b71 ("Add "--show-all" revision walker flag for debugging",
2008-02-13). In the latter commit, the commit author seems to be
deciding between not showing an unparsed commit at all and showing an
unparsed commit without the message (which is what the commit does), and
did not mention parsing the unparsed commit, so I prefer to preserve the
existing behavior.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Introduce fetch-object, providing the ability to fetch one object from a
promisor remote.
This uses fetch-pack. To do this, the transport mechanism has been
updated with 2 flags, "from-promisor" to indicate that the resulting
pack comes from a promisor remote (and thus should be annotated as such
by index-pack), and "no-dependents" to indicate that only the objects
themselves need to be fetched (but fetching additional objects is
nevertheless safe).
Whenever "no-dependents" is used, fetch-pack will refrain from using any
object flags, because it is most likely invoked as part of a dynamic
object fetch by another Git command (which may itself use object flags).
An alternative to this is to leave fetch-pack alone, and instead update
the allocation of flags so that fetch-pack's flags never overlap with
any others, but this will end up shrinking the number of flags available
to nearly every other Git command (that is, every Git command that
accesses objects), so the approach in this commit was used instead.
This will be tested in a subsequent commit.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In b495697b82 (fetch-pack: avoid repeatedly re-scanning pack
directory, 2013-01-26), we noticed that everything_local()
could waste time trying to find and parse objects which we
_expect_ to be missing. The solution was to put
has_sha1_file() in front of parse_object() to skip the
more-expensive parse attempt.
That optimization was negated later when has_sha1_file()
learned to do the same re-scan in 45e8a74873 (has_sha1_file:
re-check pack directory before giving up, 2013-08-30).
We can restore it by using the "quick" flag to tell
has_sha1_file (actually has_object_file these days) that we
prefer speed to thoroughness for this call. See also the
fixes in 5827a0354 and 0eeb077be7 for prior art and
discussion on using the "quick" flag for these cases.
The recently-added performance regression test in p5551
demonstrates the problem. You can see the original fix:
Test b495697b82^ b495697b82
--------------------------------------------------------
5551.4: fetch 1.68(1.33+0.35) 0.87(0.69+0.18) -48.2%
and then the regression:
Test 45e8a74873^ 45e8a74873
---------------------------------------------------------
5551.4: fetch 0.96(0.77+0.19) 2.55(2.04+0.50) +165.6%
and now our fix:
Test HEAD^ HEAD
--------------------------------------------------------
5551.4: fetch 7.21(6.58+0.63) 5.47(5.04+0.43) -24.1%
You can also see that other things have gotten a lot slower
since 2013. We'll deal with those in separate patches.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This is similar to using the hashmap in hashmap.c, but with an
easier-to-use API. In particular, custom entry comparisons no longer
need to be written, and lookups can be done without constructing a
temporary entry structure.
This is implemented as a thin wrapper over the hashmap API. In
particular, this means that there is an additional 4-byte overhead due
to the fact that the first 4 bytes of the hash is redundantly stored.
For now, I'm taking the simpler approach, but if need be, we can
reimplement oidmap without affecting the callers significantly.
oidset has been updated to use oidmap.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Masaya Suzuki