IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
A talloc pool is a chunk of memory that can be used as a context for further
talloc calls. Allocations with the pool as the parent just chew from that
memory by incrementing a pointer. If the talloc pool is full, then we fall back
to the normal system-level malloc(3) to get memory.
The use case for talloc pools is the transient memory that is used for handling
a single SMB request. Incrementing a pointer will be way faster than any malloc
implementation.
There is a downside of this: If you use talloc_steal() to move something out of
the pool, the whole pool memory is kept around until the last object inside the
pool is freed. So if you talloc_free() the pool, it might happen that the
memory is freed later. So don't hang anything off a talloc pool that should
live long.
Volker
cost of talloc versus malloc. The size parameter in the test suite was
constantly increasing, leading to a worst case for malloc. It is far
more common to have talloc calls of 100 bytes or lower, so change the
benchmark to reflect this.
This makes talloc look much worse - on my laptop I now get:
talloc: 5615164 ops/sec
malloc: 14337130 ops/sec
I'm working on improving that.
'subunit' protocol. This allows us to easily plug EJS scripts or binaries that
can't depend on -ltorture into smbtorture. The protocol is very simple:
- write "comments" to stderr
Example output on stdout:
test: foo
success: foo
test: bar
success: bar
test: blah
failure: blah [
dummy.c:30: Expression 1 != 2 failed!
]
test: blie
skip: blie [
Iconv support not built in
]
I've already converted the talloc testsuite.
output in the testsuite rather than just True or False for a
set of tests.
The aim is to use this for:
* known failure lists (run all tests and detect tests that
started working or started failing). This
would allow us to get rid of the RPC-SAMBA3-* tests
* nicer torture output
* simplification of the testsuite system
* compatibility with other unit testing systems
* easier usage of smbtorture (being able to run one test
and automatically set up the environment for that)
This is still a work-in-progress; expect more updates over the next couple of
days.
- talloc_free(talloc_autofree_context()); should not result
in a SIGABORT on exit
- add a test for this, but this test can also pass in the standalone build
and samba3, as samba4 uses talloc_autofree_context()
metze
meant for moving pointers between structures. The difference is that
talloc_move() will zero the source pointer, thus ensuring you don't
reference the pointer in the old context.
talloc_move() is appropriate in some, but not all cases where we use
talloc_steal() now. The interface came out of a discussion with
Jeremy.
isolation of our portability environment from the main code, and also
simplifies the includes system (no separate #ifdef _SAMBA_BUILD for
tdb. ldb etc now)
type safe. This only works on recent gcc versions. With other
compilers it reverts to a non-typesafe cast
The patch also ensures that talloc_free() does not change error on
systems where free() can change errno
have a structure that references one of its parents, and a parent of
that parent is freed, then the whole structure should be freed, not
just the reference.
this was found by the change notify code, as a side effect of fixing
the memory leak yesterday
the type names that talloc already keeps around for pointers, and
allows the user to type check void* private pointers. It can also be
used to implement polymorphism in C, as I'm sure someone would have
pointed out to me sooner or later :-)
- expanded the talloc_realloc() test a little (I was concerned about a
possible bug, which turned out to be an illusion)
- don't enable gcov by default in Makefile.talloc
