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.
This allows us to re-initialise a tevent context without destroying
the pointer. That means that if someone keeps a long term ptr to the
event context across a fork it will still work.
This also brings the memory handling in single and standard process
models much closer together, which means less bugs that we don't find
with make test.
Revert 23abcd2318c69753aa2a144e1dc0f9cf9efdb705 and fix logic bug.
The current code loops through the event contexts, when it sees a different
one, it notifies the current one (ev) and updates ev to point to the new one.
This is dumb, because:
(1) ev starts as NULL, so this code crashes, and
(2) The final context will not be notified.
The correct fix for this is to update ev to the new one, then notify it.
Volker's fix works because we currently always have one event context.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
USAGE: abi_checks.sh LIBRARY_NAME header1 [header2 ...]
This creates symbol signature lists using the mksyms and mksigs scripts
and compares them with the checked in lists.
Michael
This produces output like the output gcc produces when
invoked with the -aux-info switch.
Run like this: cat include/tevent.h | ./script/mksigs.pl
This simple parser is probably too coarse to handle all
possible header files, but it treats tevent.h correctly...
Michael
SA_INFO_QUEUE_COUNT *MUST* be a power of 2, in order for the ring buffer
wrap to work correctly at the 32 bit boundary. Thanks to Petr
Vandrovec <petr@vandrovec.name> for this.
When the first signal arrives, tevent_common_signal_handler() crashed: "ev" is
initialized to NULL, so the first "write(ev->pipe_fds[1], &c, 1);" dereferences
NULL.
Rusty, Tridge, please check. Also, can you tell me a bit more about the
environment you tested this in? I'd be curious to see where this survived.
Thanks,
Volker
The "hack_fds" were never closed before; now they're inside event_context
they should be closed when that is destroyed.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
I don't know if this is a problem in real life.
The code assumes there's only one tevent_context; all signals will notify
the first event context. That's counter-intuitive if you ever use more
than one, and there's nothing else in this code which prevents it AFAICT.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We carefully preserve the old signal handler, but we replace it before
we've set up everything; in particular, if we fail setting up the
pipe_hack we could write a NUL char to stdout (fd 0), instead of
calling the old signal handler.
Replace the signal handler as the very last thing we do.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
To be completely honest, I don't quite know whether to laugh or cry at
this one:
1 + (0xFFFFFFFF & ~(s.seen - s.count))
== 1 + (~(s.seen - s.count)) # s.seen, s.count are uint32_t
== s.count - s.seen # -A == ~A + 1
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Based on a patch submitted by Petr Vandrovec <petr@vandrovec.name>.
Multiple pending signals with siginfo_t's weren't being handled correctly
leading to smbd abort with kernel oplock signals.
Jeremy
