Bundle URIs part 4.
* ds/bundle-uri-4:
clone: unbundle the advertised bundles
bundle-uri: download bundles from an advertised list
bundle-uri: allow relative URLs in bundle lists
strbuf: introduce strbuf_strip_file_from_path()
bundle-uri: serve bundle.* keys from config
bundle-uri client: add helper for testing server
transport: rename got_remote_heads
bundle-uri client: add boolean transfer.bundleURI setting
clone: request the 'bundle-uri' command when available
t: create test harness for 'bundle-uri' command
protocol v2: add server-side "bundle-uri" skeleton
Bundle providers may want to distribute that data across multiple CDNs.
This might require a change in the base URI, all the way to the domain
name. If all bundles require an absolute URI in their 'uri' value, then
every push to a CDN would require altering the table of contents to
match the expected domain and exact location within it.
Allow a bundle list to specify a relative URI for the bundles. This URI
is based on where the client received the bundle list. For a list
provided in the 'bundle-uri' protocol v2 command, the Git remote URI is
the base URI. Otherwise, the bundle list was provided from an HTTP URI
not using the Git protocol, and that URI is the base URI. This allows
easier distribution of bundle data.
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a 'test-tool bundle-uri ls-remote' command. This is a thin wrapper
for issuing protocol v2 "bundle-uri" commands to a server, and to the
parsing routines in bundle-uri.c.
In the "git clone" case we'll have already done the handshake(),
but not here. Add an extra case to check for this handshake in
get_bundle_uri() for ease of use for future callers.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The 'got_remote_heads' member of 'struct git_transport_data' was used
historically to indicate that the initial server connection was made and
the ref advertisement was returned. With protocol v2, that initial
handshake does not necessarily include the ref advertisement, so this
member is not an accurate name. Thankfully, all uses of the member are
only checking to see if the handshake should take place, not whether or
not some local data has the ref advertisement.
Rename the member to 'finished_handshake' to represent the proper state.
Note that the variable is only set to 1 during the handshake() method.
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The yet-to-be introduced client support for bundle-uri will always
fall back on a full clone, but we'd still like to be able to ignore a
server's bundle-uri advertisement entirely.
The new transfer.bundleURI config option defaults to 'false', but a user
can set it to 'true' to enable checking for bundle URIs from the origin
Git server using protocol v2.
Co-authored-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Set up all the needed client parts of the 'bundle-uri' protocol v2
command, without actually doing anything with the bundle URIs.
If the server says it supports 'bundle-uri' teach Git to issue the
'bundle-uri' command after the 'ls-refs' during 'git clone'. The
returned key=value pairs are passed to the bundle list code which is
tested using a different ingest mechanism in t5750-bundle-uri-parse.sh.
At this point, Git does nothing with that bundle list. It will not
download any of the bundles. That will come in a later change after
these protocol bits are finalized.
The no-op client is initially used only by 'git clone' to test the basic
functionality, and eventually will bootstrap the initial download of Git
objects during a fresh clone. The bundle URI client will not be
integrated into other fetches until a mechanism is created to select a
subset of bundles for download.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Define the logical elements of a "bundle list", data structure to
store them in-core, format to transfer them, and code to parse
them.
* ds/bundle-uri-3:
bundle-uri: suppress stderr from remote-https
bundle-uri: quiet failed unbundlings
bundle: add flags to verify_bundle()
bundle-uri: fetch a list of bundles
bundle: properly clear all revision flags
bundle-uri: limit recursion depth for bundle lists
bundle-uri: parse bundle list in config format
bundle-uri: unit test "key=value" parsing
bundle-uri: create "key=value" line parsing
bundle-uri: create base key-value pair parsing
bundle-uri: create bundle_list struct and helpers
bundle-uri: use plain string in find_temp_filename()
The verify_bundle() method has a 'verbose' option, but we will want to
extend this method to have more granular control over its output. First,
replace this 'verbose' option with a new 'flags' option with a single
possible value: VERIFY_BUNDLE_VERBOSE.
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
An earlier patch discussed and fixed a scenario where Git could be used
as a vector to exfiltrate sensitive data through a Docker container when
a potential victim clones a suspicious repository with local submodules
that contain symlinks.
That security hole has since been plugged, but a similar one still
exists. Instead of convincing a would-be victim to clone an embedded
submodule via the "file" protocol, an attacker could convince an
individual to clone a repository that has a submodule pointing to a
valid path on the victim's filesystem.
For example, if an individual (with username "foo") has their home
directory ("/home/foo") stored as a Git repository, then an attacker
could exfiltrate data by convincing a victim to clone a malicious
repository containing a submodule pointing at "/home/foo/.git" with
`--recurse-submodules`. Doing so would expose any sensitive contents in
stored in "/home/foo" tracked in Git.
For systems (such as Docker) that consider everything outside of the
immediate top-level working directory containing a Dockerfile as
inaccessible to the container (with the exception of volume mounts, and
so on), this is a violation of trust by exposing unexpected contents in
the working copy.
To mitigate the likelihood of this kind of attack, adjust the "file://"
protocol's default policy to be "user" to prevent commands that execute
without user input (including recursive submodule initialization) from
taking place by default.
Suggested-by: Jeff King <peff@peff.net>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
A couple of bugfixes with code clean-up.
* jk/list-objects-filter-cleanup:
list-objects-filter: convert filter_spec to a strbuf
list-objects-filter: add and use initializers
list-objects-filter: handle null default filter spec
list-objects-filter: don't memset after releasing filter struct
The code that manages list-object-filter structure, used in partial
clones, leaked the instances, which has been plugged.
* jk/plug-list-object-filter-leaks:
prepare_repo_settings(): plug leak of config values
list_objects_filter_options: plug leak of filter_spec strings
transport: free filter options in disconnect_git()
transport: deep-copy object-filter struct for fetch-pack
list_objects_filter_copy(): deep-copy sparse_oid_name field
In 7e2619d8ff (list_objects_filter_options: plug leak of filter_spec
strings, 2022-09-08), we noted that the filter_spec string_list was
inconsistent in how it handled memory ownership of strings stored in the
list. The fix there was a bit of a band-aid to set the "strdup_strings"
variable right before adding anything.
That works OK, and it lets the users of the API continue to
zero-initialize the struct. But it makes the code a bit hard to follow
and accident-prone, as any other spots appending the filter_spec need to
think about whether to set the strdup_strings value, too (there's one
such spot in partial_clone_get_default_filter_spec(), which is probably
a possible memory leak).
So let's do that full cleanup now. We'll introduce a
LIST_OBJECTS_FILTER_INIT macro and matching function, and use them as
appropriate (though it is for the "_options" struct, this matches the
corresponding list_objects_filter_release() function).
This is harder than it seems! Many other structs, like
git_transport_data, embed the filter struct. So they need to initialize
it themselves even if the rest of the enclosing struct is OK with
zero-initialization. I found all of the relevant spots by grepping
manually for declarations of list_objects_filter_options. And then doing
so recursively for structs which embed it, and ones which embed those,
and so on.
I'm pretty sure I got everything, but there's no change that would alert
the compiler if any topics in flight added new declarations. To catch
this case, we now double-check in the parsing function that things were
initialized as expected and BUG() if appropriate.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If a user of the transport API calls transport_set_option() with
TRANS_OPT_LIST_OBJECTS_FILTER, it doesn't pass a struct, but rather a
string with the filter-spec, which the transport code then stores in its
own list_objects_filter_options struct.
When the caller is done and we call transport_disconnect(), the contents
of that filter struct are then leaked. We should release it before
freeing the transport struct.
Another way to solve this would be for transport_set_option() to pass a
pointer to the struct. But that's awkward, because there's a generic
transport-option interface that always takes a string. Plus it opens up
questions of memory lifetimes; by storing its own filter-options struct,
the transport code remains self-contained.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When the transport code for the git protocol calls into fetch_pack(), it
has to fill out a fetch_pack_args struct that is mostly taken from the
transport options. We pass along any object-filter data by doing a
struct assignment of the list_objects_filter_options struct. But doing
so isn't safe; it contains allocated pointers in its filter_spec
string_list, which could lead to a double-free if one side mutates or
frees the string_list.
And indeed, the fetch-pack code does clear and rewrite the list via
expand_list_objects_filter_spec(), leaving the transport code with
dangling pointers.
This hasn't been a problem so far, though, because the transport code
doesn't look further at the filter struct. But it should, because in
some cases (when fetch-pack doesn't rewrite the list), it ends up
leaking the string_list.
So let's start by turning this shallow copy into a deep one, which
should let us fix the transport leak in a subsequent patch. Likewise,
we'll free the deep copy we made here when we're done with it (to avoid
leaking).
Note that it would also work to pass fetch-pack a pointer to our filter
struct, rather than a copy. But it's awkward for fetch-pack to take a
pointer in its arg struct; the actual git-fetch-pack command allocates a
fetch_pack_args struct on the stack and expects it to contain the filter
options. It could be rewritten to avoid this, but a deep copy serves our
purposes just as well.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
As reported in [1] the "UNUSED(var)" macro introduced in
2174b8c75de (Merge branch 'jk/unused-annotation' into next,
2022-08-24) breaks coccinelle's parsing of our sources in files where
it occurs.
Let's instead partially go with the approach suggested in [2] of
making this not take an argument. As noted in [1] "coccinelle" will
ignore such tokens in argument lists that it doesn't know about, and
it's less of a surprise to syntax highlighters.
This undoes the "help us notice when a parameter marked as unused is
actually use" part of 9b24034754 (git-compat-util: add UNUSED macro,
2022-08-19), a subsequent commit will further tweak the macro to
implement a replacement for that functionality.
1. https://lore.kernel.org/git/220825.86ilmg4mil.gmgdl@evledraar.gmail.com/
2. https://lore.kernel.org/git/220819.868rnk54ju.gmgdl@evledraar.gmail.com/
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
get_refs_from_bundle() is a virtual function which must match the
signature of other transports, but it doesn't look at its
transport_options at all. This isn't a bug, because not all transports
necessarily support all options. Let's mark it as unused to appease
-Wunused-parameter.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The word "whitelist" has cultural implications that are not inclusive.
Thankfully, it is not difficult to reword and avoid its use.
The GIT_ALLOW_PROTOCOL environment variable was referred to as a
"whitelist", but the word "allow" is already part of the variable.
Replace "whitelist" with "allow_list" in these cases to demonstrate that
we are processing a list of allowed protocols.
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Leakfix.
* fh/transport-push-leakfix:
transport: free local and remote refs in transport_push()
transport: unify return values and exit point from transport_push()
transport: remove unnecessary indenting in transport_push()
Fix memory leaks in transport_push(), where remote_refs and local_refs
are never freed.
116 bytes in 1 blocks are definitely lost in loss record 56 of 103
at 0x484486F: malloc (vg_replace_malloc.c:381)
by 0x4938D7E: strdup (strdup.c:42)
by 0x628418: xstrdup (wrapper.c:39)
by 0x4FD454: process_capabilities (connect.c:232)
by 0x4FD454: get_remote_heads (connect.c:354)
by 0x610A38: handshake (transport.c:333)
by 0x612B02: transport_push (transport.c:1302)
by 0x4803D6: push_with_options (push.c:357)
by 0x4811D6: do_push (push.c:414)
by 0x4811D6: cmd_push (push.c:650)
by 0x405210: run_builtin (git.c:465)
by 0x405210: handle_builtin (git.c:719)
by 0x406363: run_argv (git.c:786)
by 0x406363: cmd_main (git.c:917)
by 0x404F17: main (common-main.c:56)
5,912 (388 direct, 5,524 indirect) bytes in 2 blocks are definitely lost in loss record 98 of 103
at 0x4849464: calloc (vg_replace_malloc.c:1328)
by 0x628705: xcalloc (wrapper.c:150)
by 0x5C216D: alloc_ref_with_prefix (remote.c:975)
by 0x5C232A: alloc_ref (remote.c:983)
by 0x5C232A: one_local_ref (remote.c:2299)
by 0x5C232A: one_local_ref (remote.c:2289)
by 0x5BDB03: do_for_each_repo_ref_iterator (iterator.c:418)
by 0x5B4C4F: do_for_each_ref (refs.c:1486)
by 0x5B4C4F: refs_for_each_ref (refs.c:1492)
by 0x5B4C4F: for_each_ref (refs.c:1497)
by 0x5C6ADF: get_local_heads (remote.c:2310)
by 0x612A85: transport_push (transport.c:1286)
by 0x4803D6: push_with_options (push.c:357)
by 0x4811D6: do_push (push.c:414)
by 0x4811D6: cmd_push (push.c:650)
by 0x405210: run_builtin (git.c:465)
by 0x405210: handle_builtin (git.c:719)
by 0x406363: run_argv (git.c:786)
by 0x406363: cmd_main (git.c:917)
Signed-off-by: Frantisek Hrbata <frantisek@hrbata.com>
Reviewed-by: Josh Steadmon <steadmon@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It seems there is no reason to return 1 instead of -1 when push_refs()
is not set in transport vtable. Let's unify the error return values and
use the done label as a single exit point from transport_push().
Suggested-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Frantisek Hrbata <frantisek@hrbata.com>
Reviewed-by: Josh Steadmon <steadmon@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Remove the big indented block for transport_push() check in transport vtable
and let's just return error immediately. Hopefully this makes the code
more readable.
Signed-off-by: Frantisek Hrbata <frantisek@hrbata.com>
Reviewed-by: Josh Steadmon <steadmon@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Introduce and apply coccinelle rule to discourage an explicit
comparison between a pointer and NULL, and applies the clean-up to
the maintenance track.
* ep/maint-equals-null-cocci:
tree-wide: apply equals-null.cocci
tree-wide: apply equals-null.cocci
contrib/coccinnelle: add equals-null.cocci
"git fetch --refetch" learned to fetch everything without telling
the other side what we already have, which is useful when you
cannot trust what you have in the local object store.
* rc/fetch-refetch:
docs: mention --refetch fetch option
fetch: after refetch, encourage auto gc repacking
t5615-partial-clone: add test for fetch --refetch
fetch: add --refetch option
builtin/fetch-pack: add --refetch option
fetch-pack: add refetch
fetch-negotiator: add specific noop initializer
Teach fetch and transports the --refetch option to force a full fetch
without negotiating common commits with the remote. Use when applying a
new partial clone filter to refetch all matching objects.
Signed-off-by: Robert Coup <robert@coup.net.nz>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Amend the logic added in fddf2ebe38 (transport: teach all vtables to
allow fetch first, 2019-08-21) and save ourselves pointless work in
fetch_refs_from_bundle().
The fetch_refs_from_bundle() caller doesn't care about the "struct
ref *result" return value of get_refs_from_bundle(), and doesn't need
any of the work we were doing in looping over the
"data->header.references" in get_refs_from_bundle().
So this change saves us work, and also fixes a memory leak that we had
when called from fetch_refs_from_bundle(). The other caller of
get_refs_from_bundle() is the "get_refs_list" member we set up for the
"struct transport_vtable bundle_vtable". That caller does care about
the "struct ref *result" return value.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix a memory leak in codepaths that use the "struct
transport_ls_refs_options" API. Since the introduction of the struct
in 39835409d1 (connect, transport: encapsulate arg in struct,
2021-02-05) the caller has been responsible for freeing it.
That commit in turn migrated code originally added in
402c47d939 (clone: send ref-prefixes when using protocol v2,
2018-07-20) and b4be74105f (ls-remote: pass ref prefixes when
requesting a remote's refs, 2018-03-15). Only some of those codepaths
were releasing the allocated resources of the struct, now all of them
will.
Mark the "t/t5511-refspec.sh" test as passing when git is compiled
with SANITIZE=leak. They'll now be listed as running under the
"GIT_TEST_PASSING_SANITIZE_LEAK=true" test mode (the "linux-leaks" CI
target). Previously 24/47 tests would fail.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Some lockfile code called free() in signal-death code path, which
has been corrected.
* ps/lockfile-cleanup-fix:
fetch: fix deadlock when cleaning up lockfiles in async signals
When fetching packfiles, we write a bunch of lockfiles for the packfiles
we're writing into the repository. In order to not leave behind any
cruft in case we exit or receive a signal, we register both an exit
handler as well as signal handlers for common signals like SIGINT. These
handlers will then unlink the locks and free the data structure tracking
them. We have observed a deadlock in this logic though:
(gdb) bt
#0 __lll_lock_wait_private () at ../sysdeps/unix/sysv/linux/x86_64/lowlevellock.S:95
#1 0x00007f4932bea2cd in _int_free (av=0x7f4932f2eb20 <main_arena>, p=0x3e3e4200, have_lock=0) at malloc.c:3969
#2 0x00007f4932bee58c in __GI___libc_free (mem=<optimized out>) at malloc.c:2975
#3 0x0000000000662ab1 in string_list_clear ()
#4 0x000000000044f5bc in unlock_pack_on_signal ()
#5 <signal handler called>
#6 _int_free (av=0x7f4932f2eb20 <main_arena>, p=<optimized out>, have_lock=0) at malloc.c:4024
#7 0x00007f4932bee58c in __GI___libc_free (mem=<optimized out>) at malloc.c:2975
#8 0x000000000065afd5 in strbuf_release ()
#9 0x000000000066ddb9 in delete_tempfile ()
#10 0x0000000000610d0b in files_transaction_cleanup.isra ()
#11 0x0000000000611718 in files_transaction_abort ()
#12 0x000000000060d2ef in ref_transaction_abort ()
#13 0x000000000060d441 in ref_transaction_prepare ()
#14 0x000000000060e0b5 in ref_transaction_commit ()
#15 0x00000000004511c2 in fetch_and_consume_refs ()
#16 0x000000000045279a in cmd_fetch ()
#17 0x0000000000407c48 in handle_builtin ()
#18 0x0000000000408df2 in cmd_main ()
#19 0x00000000004078b5 in main ()
The process was killed with a signal, which caused the signal handler to
kick in and try free the data structures after we have unlinked the
locks. It then deadlocks while calling free(3P).
The root cause of this is that it is not allowed to call certain
functions in async-signal handlers, as specified by signal-safety(7).
Next to most I/O functions, this list of disallowed functions also
includes memory-handling functions like malloc(3P) and free(3P) because
they may not be reentrant. As a result, if we execute such functions in
the signal handler, then they may operate on inconistent state and fail
in unexpected ways.
Fix this bug by not calling non-async-signal-safe functions when running
in the signal handler. We're about to re-raise the signal anyway and
will thus exit, so it's not much of a problem to keep the string list of
lockfiles untouched. Note that it's fine though to call unlink(2), so
we'll still clean up the lockfiles correctly.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Reviewed-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change a pattern of hardcoding an "argv" array size, populating it and
assigning to the "argv" member of "struct child_process" to instead
use "strvec_push()" to add data to the "args" member.
As noted in the preceding commit this moves us further towards being
able to remove the "argv" member in a subsequent commit
These callers could have used strvec_pushl(), but moving to
strvec_push() makes the diff easier to read, and keeps the arguments
aligned as before.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Move the find_hook() function from run-command.c to a new hook.c
library. This change establishes a stub library that's pretty
pointless right now, but will see much wider use with Emily Shaffer's
upcoming "configuration-based hooks" series.
Eventually all the hook related code will live in hook.[ch]. Let's
start that process by moving the simple find_hook() function over
as-is.
Signed-off-by: Emily Shaffer <emilyshaffer@google.com>
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Code clean-up around "git serve".
* ab/serve-cleanup:
upload-pack: document and rename --advertise-refs
serve.[ch]: remove "serve_options", split up --advertise-refs code
{upload,receive}-pack tests: add --advertise-refs tests
serve.c: move version line to advertise_capabilities()
serve: move transfer.advertiseSID check into session_id_advertise()
serve.[ch]: don't pass "struct strvec *keys" to commands
serve: use designated initializers
transport: use designated initializers
transport: rename "fetch" in transport_vtable to "fetch_refs"
serve: mark has_capability() as static
Since the "flags" parameter was added in be042aff24 (Teach progress
eye-candy to fetch_refs_from_bundle(), 2011-09-18) there's never been
more than the one flag: BUNDLE_VERBOSE.
Let's have the only caller who cares about that pass "-v" itself
instead through new "extra_index_pack_args" parameter. The flexibility
of being able to pass arbitrary arguments to "unbundle" will be used
in a subsequent commit.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change the assignments to the various transport_vtables to use
designated initializers, this makes the code easier to read and
maintain.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Rename the "fetch" member of the transport_vtable to "fetch_refs" for
consistency with the existing "push_refs". Neither of them just push
"refs" but refs and objects, but having the two match makes the code
more readable than having it be inconsistent, especially since
"fetch_refs" is a lot easier to grep for than "fetch".
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Code cleanup around struct_type_init() functions.
* ab/struct-init:
string-list.h users: change to use *_{nodup,dup}()
string-list.[ch]: add a string_list_init_{nodup,dup}()
dir.[ch]: replace dir_init() with DIR_INIT
*.c *_init(): define in terms of corresponding *_INIT macro
*.h: move some *_INIT to designated initializers
Move away from the "struct ref_list" in bundle.c in favor of the
almost identical string-list.c API.
That API fits this use-case perfectly, but did not exist in its
current form when this code was added in 2e0afafebd (Add git-bundle:
move objects and references by archive, 2007-02-22), with hindsight we
could have used the path-list API, which later got renamed to
string-list. See 8fd2cb4069 (Extract helper bits from
c-merge-recursive work, 2006-07-25)
We need to change "name" to "string" and "oid" to "util" to make this
conversion, but other than that the APIs are pretty much identical for
what bundle.c made use of.
Let's also replace the memset(..,0,...) pattern with a more idiomatic
"INIT" macro, and finally add a *_release() function so to free the
allocated memory.
Before this the add_to_ref_list() would leak memory, now e.g. "bundle
list-heads" reports no memory leaks at all under valgrind.
In the bundle_header_init() function we're using a clever trick to
memcpy() what we'd get from the corresponding
BUNDLE_HEADER_INIT. There is a concurrent series to make use of that
pattern more generally, see [1].
1. https://lore.kernel.org/git/cover-0.5-00000000000-20210701T104855Z-avarab@gmail.com/
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Acked-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In preparation for moving away from accessing the OID and name via the
"oid" and "name" slots in a subsequent commit, change the code that
accesses it to use named variables. This makes the subsequent change
smaller.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Acked-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change all in-tree users of the string_list_init(LIST, BOOL) API to
use string_list_init_{nodup,dup}(LIST) instead.
As noted in the preceding commit let's leave the now-unused
string_list_init() wrapper in-place for any in-flight users, it can be
removed at some later date.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When fetching with the v0 protocol over ssh (or a local upload-pack with
pipes), the server closes the connection as soon as it is finished
sending the pack. So even though the client may still be operating on
the data via index-pack (e.g., resolving deltas, checking connectivity,
etc), the server has released all resources.
With the v2 protocol, however, the server considers the ssh session only
as a transport, with individual requests coming over it. After sending
the pack, it goes back to its main loop, waiting for another request to
come from the client. As a result, the ssh session hangs around until
the client process ends, which may be much later (because resolving
deltas, etc, may consume a lot of CPU).
This is bad for two reasons:
- it's consuming resources on the server to leave open a connection
that won't see any more use
- if something bad happens to the ssh connection in the meantime (say,
it gets killed by the network because it's idle, as happened in a
real-world report), then ssh will exit non-zero, and we'll propagate
the error up the stack.
The server is correct here not to hang up after serving the pack. The v2
protocol's design is meant to allow multiple requests like this, and
hanging up would be the wrong thing for a hypothetical client which was
planning to make more requests (though in practice, the git.git client
never would, and I doubt any other implementations would either).
The right thing is instead for the client to signal to the server that
it's not interested in making more requests. We can do that by closing
the pipe descriptor we use to write to ssh. This will propagate to the
server upload-pack as an EOF when it tries to read the next request (and
then it will close its half, and the whole connection will go away).
It's important to do this "half duplex" shutdown, because we have to do
it _before_ we actually receive the pack. This is an artifact of the way
fetch-pack and index-pack (or unpack-objects) interact. We hand the
connection off to index-pack (really, a sideband demuxer which feeds
it), and then wait until it returns. And it doesn't do that until it has
resolved all of the deltas in the pack, even though it was done reading
from the server long before.
So just closing the connection fully after index-pack returns would be
too late; we'd have held it open much longer than was necessary. And
teaching index-pack to close the connection is awkward. It's not even
seeing the whole conversation (the sideband demuxer is, but it doesn't
actually know what's in the packets, or when the end comes).
Note that this close() is happening deep within the transport code. It's
possible that a caller would want to perform other operations over the
same ssh transport after receiving the pack. But as of the current code,
none of the callers do, and there haven't been discussions of any plans
to change this. If we need to support that later, we can probably do so
by passing down a flag for "you're the last request on the transport;
it's OK to close" instead of the code just assuming that's true.
The description above all discusses v2 ssh, so it's worth thinking about
how this interacts with other protocols:
- in v0 protocols, we could do the same half-duplex shutdown (it just
goes into the v0 do_fetch_pack() instead). This does work, but since
it doesn't have the same persistence problem in the first place,
there's little reason to change it at this point.
- local fetches against git-upload-pack on the same machine will
behave the same as ssh (they are talking over two pipes, and see EOF
on their input pipe)
- fetches against git-daemon will run this same code, and close one of
the descriptors. In practice, this won't do anything, since there
our two descriptors are dups of each other, and not part of a
half-duplex pair. The right thing would probably be to call
shutdown(SHUT_WR) on it. I didn't bother with that here. It doesn't
face the same error-code problem (since it's just a TCP connection),
so it's really only an optimization problem. And git:// is not that
widely used these days, and has less impact on server resources than
an ssh termination.
- v2 http doesn't suffer from this problem in the first place, as our
pipes terminate at a local git-remote-https, which is passing data
along as individual requests via curl. Probably curl is keeping the
TCP/TLS connection open for more requests, and we might be able to
tell it manually "hey, we are done making requests now". But I think
that's much less important. It again doesn't suffer from the
error-code problem, and HTTP keepalive is pretty well understood
(importantly, the timeouts can be set low, because clients like curl
know how to reconnect for subsequent requests if necessary). So it's
probably not worth figuring out how to tell curl that we're done
(though if we do, this patch is probably the first step anyway;
fetch-pack closes the pipe back to remote-https, which would be the
signal that it should tell curl we're done).
The code is pretty straightforward. We close the pipe at the right
moment, and set it to -1 to mark it as invalid. I modified the later
cleanup code to avoid calling close(-1). That's not strictly necessary,
since close(-1) is a noop, but hopefully makes things a bit more obvious
to a reader.
I suspect that trying to call more transport functions after the close()
(e.g., calling transport_fetch_refs() again) would fail, as it's not
smart enough to realize we need to re-open the ssh connection. But
that's already true when v0 is in use. And no current callers want to do
that (and again, the solution is probably a flag in the transport code
to keep things open, which can be added later).
There's no test here, as the situation it covers is inherently racy (the
question is when upload-pack exits, compared to when index-pack finishes
resolving deltas and exits). The rather gross shell snippet below does
recreate the problematic situation; when run on a sufficiently-large
repository (git.git works fine), it kills an "idle" upload-pack while
the client is resolving deltas, leading to a failed clone.
(
git clone --no-local --progress . foo.git 2>&1
echo >&2 "clone exit code=$?"
) |
tr '\r' '\n' |
while read line
do
case "$done,$line" in
,Resolving*)
echo "hit resolving deltas; killing upload-pack"
killall -9 git-upload-pack
done=t
;;
esac
done
Reported-by: Greg Pflaum <greg.pflaum@pnp-hcl.com>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git push" learns to discover common ancestor with the receiving
end over protocol v2.
* jt/push-negotiation:
send-pack: support push negotiation
fetch: teach independent negotiation (no packfile)
fetch-pack: refactor command and capability write
fetch-pack: refactor add_haves()
fetch-pack: refactor process_acks()
