Peter Stuge [Sat, 7 Apr 2012 03:20:36 +0000 (05:20 +0200)]
examples: Use snprintf() instead of sprintf() in dpfp and dpfp_threaded
The OpenBSD C compiler generates a warning for every use of sprintf()
and for a good reason. Reported in http://marc.info/?m=
133376187514495
Reported-by: Xiaofan Chen <xiaofanc@gmail.com>
Peter Stuge [Sat, 7 Apr 2012 01:55:36 +0000 (03:55 +0200)]
OpenBSD: Fix warning: 'hpriv' may be used uninitialized
As reported in http://marc.info/?m=
133376187514495
The variable would be used uninitialized when the internal list of
open devices is empty as obsd_handle_events() checks which device
the event fd belongs to, which can not actually happen, but if it
ever does happen then we will now return LIBUSB_ERROR_NO_DEVICE.
Reported-by: Xiaofan Chen <xiaofanc@gmail.com>
Pete Batard [Mon, 9 Apr 2012 21:28:34 +0000 (22:28 +0100)]
Windows: Fix support for variadic macros
* Visual C++ did not support variadic macros until VS2005, as per
http://msdn.microsoft.com/en-us/library/ms177415%28v=VS.80%29.aspx
Ludovic Rousseau [Wed, 4 Apr 2012 07:30:12 +0000 (09:30 +0200)]
Misc: Indicate that licence is LGPL-2.1 or later
Also make the name of the licence more explicit.
Ludovic Rousseau [Wed, 4 Apr 2012 07:24:17 +0000 (09:24 +0200)]
Misc: Update list of maintainers from README
Pete Batard [Tue, 3 Apr 2012 15:31:32 +0000 (16:31 +0100)]
Misc: Fix missing libsub's -> libusbx's from previous patches
Pete Batard [Tue, 3 Apr 2012 12:42:42 +0000 (13:42 +0100)]
Doc: Update doxygen for libusbx and add logo
Pete Batard [Tue, 3 Apr 2012 12:40:39 +0000 (13:40 +0100)]
Misc: Rebrand to libusbx
* Mentions of 'libusb' in doxygen are changed to 'libusbx'
* Also update copyright notices and remove unneeded EOF LFs
Pete Batard [Mon, 2 Apr 2012 14:47:53 +0000 (15:47 +0100)]
libusbx 1.0.9
* Also updated AUTHORS and THANKS
Pete Batard [Sun, 1 Apr 2012 20:30:52 +0000 (21:30 +0100)]
Samples: Fix "warning: no newline at end of file" in xusb.c
* Also remove trailing whitespaces
Pete Batard [Sat, 31 Mar 2012 21:32:56 +0000 (22:32 +0100)]
Autotools: Fix missing DLL .def file from make dist
Pete Batard [Sat, 31 Mar 2012 21:14:47 +0000 (22:14 +0100)]
Autotools: Fix msvc dir not being included in make dist
Hans de Goede [Fri, 30 Mar 2012 12:40:32 +0000 (13:40 +0100)]
Linux: Don't spam the console with 'device unconfigured' errors
Pete Batard [Fri, 30 Mar 2012 23:17:44 +0000 (00:17 +0100)]
libusbx 1.0.9-rc5
Pete Batard [Fri, 30 Mar 2012 23:21:18 +0000 (00:21 +0100)]
Autotools: remove --enable-debug-log option in autogen.sh
Pete Batard [Fri, 30 Mar 2012 12:40:32 +0000 (13:40 +0100)]
Darwin: Fix a warning on 64 bit platforms
Pete Batard [Fri, 30 Mar 2012 12:05:12 +0000 (13:05 +0100)]
Autotools: Fix make distcheck and add ChangeLog
Pete Batard [Fri, 30 Mar 2012 12:04:18 +0000 (13:04 +0100)]
Autotools: Fix package name and URLs in configure.ac
Pete Batard [Thu, 29 Mar 2012 22:22:28 +0000 (23:22 +0100)]
libusbx 1.0.9-rc4
Pete Batard [Thu, 29 Mar 2012 22:21:25 +0000 (23:21 +0100)]
Git: add export-ignore for git config file
* Prevents release archives produced from git to contain
git related files.
Pete Batard [Wed, 28 Mar 2012 10:43:00 +0000 (11:43 +0100)]
Samples: fix a "value is always 0" Clang warning in xusb.c
* Remove the use of os_fd[i].recipient and alter comments with
regards to Windows Extended Props test limitations with WinUSB
Pete Batard [Wed, 28 Mar 2012 00:02:02 +0000 (01:02 +0100)]
Windows: fix Clang warnings about unused assignations
Pete Batard [Wed, 28 Mar 2012 09:56:32 +0000 (10:56 +0100)]
Samples: Fix missing xusb from VS project files
Pete Batard [Mon, 26 Mar 2012 22:34:25 +0000 (23:34 +0100)]
autotools: LT_LANG([Windows Resource]) makes windres check redundant
Pete Batard [Mon, 26 Mar 2012 22:48:49 +0000 (23:48 +0100)]
Examples: add xusb sample
* Also improvements to WDK build scripts:
cd to original dir on error, use 2 cores if available
Pete Batard [Mon, 26 Mar 2012 23:03:07 +0000 (00:03 +0100)]
VS2005: Bring x86_64 output directories in line with x86_32
Hans de Goede [Mon, 26 Mar 2012 22:01:48 +0000 (23:01 +0100)]
Linux: fix inaccurate comment
* We don't stop on urbs/iso pkts with less data then requested
Hans de Goede [Mon, 26 Mar 2012 22:00:28 +0000 (23:00 +0100)]
Linux: Remove unneeded status assignation
* status value was never used
Hans de Goede [Mon, 26 Mar 2012 21:54:12 +0000 (22:54 +0100)]
Linux: Translate iso pkt status codes to libusb transfer status codes
* Linux negative errno from iso pkts status must be translated and reported.
Hans de Goede [Mon, 26 Mar 2012 21:51:56 +0000 (22:51 +0100)]
Linux: Fix handling of urb status codes
* Adds EOVERFLOW handling in handle_control_completion
* Removes handling of codes that can never occur on an iso urb
Hans de Goede [Mon, 26 Mar 2012 20:37:03 +0000 (21:37 +0100)]
Linux: Don't print errors when cancel_transfer fails with NOT_FOUND
* Under some cricumstances, LIBUSB_ERROR_NOT_FOUND is an expected
return value for cancel_transfer so printing an error is undesirable.
Hans de Goede [Mon, 26 Mar 2012 20:33:39 +0000 (21:33 +0100)]
Linux: Fix cancel_transfer return value when cancelling a multi-urb transfer
* if we've successfully cancelled the last urb, and then another urb
fails with errno == EINVAL, we shouldn't return LIBUSB_ERROR_NOT_FOUND
Peter Stuge [Thu, 22 Mar 2012 00:52:59 +0000 (01:52 +0100)]
libusbi.h: Add IS_EPIN() IS_EPOUT() IS_XFERIN() and IS_XFEROUT() macros
The macros make backend code a little simpler to read, since the
transfer and endpoint direction is checked in several places across
the supported platforms.
Peter Stuge [Wed, 22 Feb 2012 13:21:54 +0000 (14:21 +0100)]
Add LIBUSB_TRANSFER_ADD_ZERO_PACKET flag to indicate need for ZLP
Some protocols which use USB require an extra zero length data packet
to signal end-of-transfer on bulk endpoints, if the last data packet
is exactly wMaxPacketSize bytes long.
This flag allows applications to inform libusb about this requirement,
so that libusb can handle the issue transparently.
At the moment the new flag is only supported on Linux, and submitting
a transfer with the flag set returns an error at submit time on other
systems. Hopefully implementations will soon follow for other systems.
References #6.
Peter Stuge [Sun, 19 Feb 2012 05:58:16 +0000 (06:58 +0100)]
libusb.h: Add comma after last value in enums to ease future expansion
Peter Stuge [Sun, 19 Feb 2012 04:39:33 +0000 (05:39 +0100)]
Linux: Rework bulk continuation check to generic kernel version check
The old check_flag_bulk_continuation() tested for one specific running
kernel version. The new kernel_version_ge() instead allows to test the
running kernel version against major, minor and sublevel parameters.
Peter Stuge [Sun, 19 Feb 2012 04:32:10 +0000 (05:32 +0100)]
OpenBSD: Correctly catch attempts to do isochronous OUT transfers
LIBUSB_ENDPOINT_OUT is the value of the direction bit, which is 0 for
OUT transfers, so the previous condition could never evaluate to true.
Peter Stuge [Sat, 18 Feb 2012 23:42:20 +0000 (00:42 +0100)]
Darwin: Do not reject devices with product id 0x0000
Fixes #128.
Peter Stuge [Mon, 13 Feb 2012 15:59:51 +0000 (16:59 +0100)]
Do not call timerfd functions when timerfd is not being used
When libusb was built with timerfd support but used on a system
without timerfd support the library would hang indefinitely on
completion of the first transfer, since timerfd functions were
being called unconditionally and the error returned when timerfd
was not being used caused a confused internal state.
Many thanks to Ivo Smits for looking into the issue, proposing
an initial solution, and helping with testing! Fixes #73.
Michael Plante [Sun, 12 Feb 2012 18:49:35 +0000 (13:49 -0500)]
Windows: Fix MSVC6 IDE warning on line continuation in libusb-1.0.rc
* libusb\libusb-1.0.rc(21): Could not find the file LIBUSB_RC.
* only shows up first time after opening workspace.
* shows up on either build or clean.
* not actually due to rc.exe, but prior to it.
* probably an IDE bug.
* does not show up when running an exported makefile.
Signed-off-by: Michael Plante <michael.plante@gmail.com>
Pete Batard [Sat, 7 Jan 2012 23:30:23 +0000 (23:30 +0000)]
Windows: Add Renesas SuperSpeed support for libusb_get_device_speed()
* tests conducted using a Renesas PCIE USB 3.0 controller and driver and
a mass storage USB 3.0 confirm that USB_NODE_CONNECTION_INFORMATION_EX
uses (undocumented) value 3 for SuperSpeed
Peter Stuge [Fri, 10 Feb 2012 17:06:43 +0000 (18:06 +0100)]
autogen.sh: Try to use libtoolize before trying glibtoolize
This helps on Mac OS X where an old glibtoolize is included in the
system and newer, manually installed, versions provide libtoolize.
See also http://marc.info/?m=
132490560131894
Peter Stuge [Wed, 8 Feb 2012 14:54:21 +0000 (15:54 +0100)]
Rename the lsusb example program to listdevs
This is intended to reduce confusion with the much more significant
lsusb utility which is part of the usbutils package.
Sean McBride [Fri, 9 Dec 2011 18:44:21 +0000 (13:44 -0500)]
core.c: Fix clang warning about useless store to a variable
Sean McBride [Fri, 9 Dec 2011 18:43:46 +0000 (13:43 -0500)]
Darwin: Clarify comments on event thread creation and termination
Sean McBride [Fri, 9 Dec 2011 18:43:46 +0000 (13:43 -0500)]
Darwin: Fix incorrect mutex protection of libusb_darwin_acfl global
Sean McBride [Fri, 9 Dec 2011 18:43:46 +0000 (13:43 -0500)]
Darwin: Use retained runloop variable instead of CFRunLoopGetCurrent()
Sean McBride [Fri, 9 Dec 2011 18:37:22 +0000 (13:37 -0500)]
Darwin: Fix libusb_init() and libusb_exit() refcount race condition
Use OSAtomicIncrement32Barrier() and OSAtomicDecrement32Barrier()
in darwin_init() and darwin_exit() to be thread safe.
Sean McBride [Fri, 9 Dec 2011 18:33:28 +0000 (13:33 -0500)]
threads_posix.c: Fix prototype warning by including threads_posix.h
Dave Camarillo [Fri, 23 Sep 2011 10:56:58 +0000 (11:56 +0100)]
Windows: Fix crash during enumeration with >= 64 connected devices
* indexes were outgrowing the unref_list array before realloc,
resulting in out of bound access and crash.
Pete Batard [Tue, 26 Oct 2010 15:31:13 +0000 (16:31 +0100)]
Windows MSVC6: Remove SetupAPI, AdvAPI32 and OLE32 dependencies
* this ensures that libusb dependent applications only need
to link with libusb on Windows
* (copied from Pete's msvc08 mods to msvc6 by Michael)
Pete Batard [Wed, 14 Dec 2011 02:18:10 +0000 (21:18 -0500)]
Windows: Output an error message on calls to libusb_get_pollfds()
Signed-off-by: Michael Plante <michael.plante@gmail.com>
Pete Batard [Fri, 4 Jun 2010 03:14:20 +0000 (22:14 -0500)]
io.c: Fix comment since pthreads isn't the only threading libusb uses
Michael Plante [Tue, 3 Aug 2010 00:59:12 +0000 (19:59 -0500)]
libusb.h: Cast in libusb_get_string_descriptor() to avoid MSVC6 warning
On MSVC6 bitwise OR promotes to int, causing the warning.
Michael Plante [Fri, 23 Dec 2011 22:50:21 +0000 (16:50 -0600)]
Windows: guid_to_string() is only discarded when macro-usbi_dbg used
* usbi_dbg encloses all references to guid_to_string
* MinGW/cygwin warn about an unused function, so the #if
squelches this warning
* MSVC6 uses a variadic function form of usbi_dbg instead
of a macro, so the compiler still "sees" guid_to_string
and it therefore needs to always be defined for MSVC6,
even if it's only a stub.
* So we define it if usbi_dbg is used OR if MSVC6 is used.
Signed-off-by: Michael Plante <michael.plante@gmail.com>
Pete Batard [Tue, 15 Nov 2011 01:11:12 +0000 (01:11 +0000)]
Windows: Fix undefined datatype 'LONG_PTR' in MSVC6
* issue reported by Elmi
Signed-off-by: Michael Plante <michael.plante@gmail.com>
Pete Batard [Sun, 13 Nov 2011 23:10:15 +0000 (23:10 +0000)]
Windows: Fix for MSVC6 preprocessor not accepting blank parameters
* unlike later iterations of Visual Studio, MSVC6 does not accept
blank parameters on macro calls [eg. CALL(a, ,b)]
* blank params were used with the DLL_DECLARE and DLL_LOAD macros
* issue reported by Elmi
Michael Plante [Tue, 26 Oct 2010 10:39:38 +0000 (11:39 +0100)]
Windows: Switch MSVC6 .dsp project files from MBCS to Unicode
* MBCS (which is different from UTF-8) only makes sense if
supporting Windows 95/98, which we don't
* (try to match Pete's vcproj changes in MSVC6)
Michael Plante [Wed, 14 Dec 2011 02:39:40 +0000 (21:39 -0500)]
Windows: Add missing MSVC6 lsusb.dsp project file
Signed-off-by: Michael Plante <michael.plante@gmail.com>
Martin Pieuchot [Mon, 30 Jan 2012 08:59:30 +0000 (09:59 +0100)]
OpenBSD backend
Xiaofan Chen [Fri, 13 Jan 2012 17:07:34 +0000 (18:07 +0100)]
configure.ac: Darwin: Move -lobjc from LIBS to PC_LIBS_PRIVATE
Since commit
40327cd134718475f6cec8935b856d4fdff2099c it is neccessary
to explicitly include -lobjc not only when linking libusb itself, but
also for programs linking statically against libusb. References #63.
See also http://marc.info/?m=
132505900202378
Xiaofan Chen [Fri, 13 Jan 2012 12:04:00 +0000 (13:04 +0100)]
examples: Link only with ../libusb/libusb-1.0.la and not with -lusb-1.0
Previous _LDFLAGS included both the freshly built libusb in ../libusb
and -lusb-1.0, where libtool would usually resolve the latter to an
already-installed libusb library in the system. The extra reference
to a second libusb library resulted in failure to build examples on
Mac OS X in some cases, and is plain wrong.
See also the thread at http://marc.info/?m=
132637593623667
Peter Stuge [Thu, 12 Jan 2012 09:34:55 +0000 (10:34 +0100)]
libusb/Makefile.am: Avoid recursive variables in .rc silent-rule
Not every make supports recursive variable expansion so some automake
versions complain about non-POSIX variable names ever since commit
70bec4a9f8ec28d36c731011fa24d37c74ad3523 which added support for
silent-rules in our rule to compile the Windows .rc file.
This commit removes the recursive variables and instead uses the
simple and generic GEN message and associated variable.
Ludovic Rousseau [Sat, 24 Sep 2011 16:27:54 +0000 (18:27 +0200)]
Core: Fix #122 warning: no previous prototype for ‘usbi_log_v’
Declare the usbi_log_v() function before using it.
Sean McBride [Fri, 14 Oct 2011 18:40:29 +0000 (14:40 -0400)]
Darwin: Set a name for the background thread on Mac OS X 10.6 and later
The call to pthread_setname_np() makes it easy to identify the
background thread in the Xcode debugger and in crash reports.
Peter Stuge [Thu, 22 Sep 2011 19:42:35 +0000 (21:42 +0200)]
libusb/Makefile.am: Add libusb-1.0.rc dependency on version.h
Peter Stuge [Thu, 22 Sep 2011 19:38:29 +0000 (21:38 +0200)]
libusb/Makefile.am: Make the libusb-1.0.rc rule support silent-rules
Peter Stuge [Thu, 22 Sep 2011 19:17:37 +0000 (21:17 +0200)]
configure.ac: Enable libtool support for Windows Resource language
Sean McBride [Mon, 19 Sep 2011 16:52:44 +0000 (12:52 -0400)]
descriptor.c: Use size_t for minor 64 bit compliance improvement
Fixes #121.
Sean McBride [Mon, 19 Sep 2011 16:42:33 +0000 (12:42 -0400)]
Darwin: Fix warnings about mismatched signed/unsigned compare
References #121.
Sean McBride [Mon, 19 Sep 2011 16:38:20 +0000 (12:38 -0400)]
Fix unused variable warnings when without timerfd and/or when on Darwin
References #121.
Ludovic Rousseau [Fri, 16 Sep 2011 16:07:56 +0000 (18:07 +0200)]
Correctly handle LIBUSB_TRANSFER_OVERFLOW in libusb_control_transfer()
sync.c: In function `libusb_control_transfer':
sync.c:122: warning: enumeration value `LIBUSB_TRANSFER_OVERFLOW' not
handled in switch
Fixes #120.
Pete Batard [Mon, 19 Sep 2011 15:01:37 +0000 (16:01 +0100)]
Windows: do not set altsetting 0 as part of libusb_release_interface()
* pointed out by Travis Robinson and Xiaofan Chen
* similar to a change advised by Alan Stern for the Linux kernel:
http://marc.info/?m=
122790204410765
Pete Batard [Mon, 19 Sep 2011 14:45:35 +0000 (15:45 +0100)]
Windows: fix communication with hubs
* issue reported by René Haunstrup in http://marc.info/?m=
130503019227814
Pete Batard [Mon, 19 Sep 2011 14:41:39 +0000 (15:41 +0100)]
Windows: add USB 3.0 controller support
* Windows 8, NEC/Renesas, TI, Fresco Logic, Etron, VIA, ASMedia
(some of which untested!)
* includes workaround for NEC/Renesas USB 3.0 root hubs
Pete Batard [Mon, 19 Sep 2011 14:28:03 +0000 (15:28 +0100)]
Windows: add static keywords and drop duplicated or unneeded code
Pete Batard [Mon, 19 Sep 2011 12:59:46 +0000 (13:59 +0100)]
Windows: fix 'windows_error_str' defined but not used when not logging
Peter Stuge [Thu, 22 Sep 2011 08:19:17 +0000 (10:19 +0200)]
libusbi.h: Silence unused context variable warnings when not logging
Pete Batard [Mon, 19 Sep 2011 12:06:22 +0000 (13:06 +0100)]
Windows: copy DLL file regardless of ddk_build.cmd dll argument case
Pete Batard [Sun, 18 Sep 2011 19:06:45 +0000 (20:06 +0100)]
Windows: Add uint32_t typedef for old MSC so that DDK can build lsusb
Vitali Lovich [Fri, 16 Sep 2011 14:53:25 +0000 (08:53 -0600)]
Darwin: fix port leak in darwin_clock_gettime()
Vitali Lovich [Fri, 16 Sep 2011 14:50:40 +0000 (08:50 -0600)]
Darwin: remove unnecessary run loop invalidation during libusb_exit()
Vitali Lovich [Fri, 16 Sep 2011 14:49:54 +0000 (08:49 -0600)]
Darwin: fix port leak during libusb_exit()
Vitali Lovich [Fri, 16 Sep 2011 14:48:28 +0000 (08:48 -0600)]
Darwin: fix potential leak on libusb_claim_interface() error
Nathan Hjelm [Fri, 16 Sep 2011 14:41:01 +0000 (08:41 -0600)]
Darwin: Fix #117 transfer size 64/32 bit issue for transfer callbacks
The transfer size is now always truncated to 32 bits.
Peter Stuge [Thu, 15 Sep 2011 15:04:43 +0000 (17:04 +0200)]
Remove username from SF FRS rsync upload command
Peter Stuge [Thu, 15 Sep 2011 14:48:35 +0000 (16:48 +0200)]
Support release candidate versions in configure.ac and libusb-1.0.rc
Peter Stuge [Thu, 15 Sep 2011 14:32:47 +0000 (16:32 +0200)]
Tidy .gitignore a little
There's no xusb, we need a glob pattern for matching .exe files, and
fpusb tarballs aren't very common in the libusb source dir.
Hans de Goede [Wed, 14 Sep 2011 08:53:48 +0000 (10:53 +0200)]
Docs: Clarify that libusb_handle_events_timeout() tv param can't be NULL
The example code and API doc for libusb_handle_events_timeout() could be
interpreted as it being OK to pass a NULL pointer for the tv argument (I
interpreted it like that when I first started coding for libusb).
This patch changes the docs to make it clear that one must always supply
a tv struct to libusb_handle_events_timeout.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Peter Stuge [Wed, 14 Sep 2011 06:23:43 +0000 (08:23 +0200)]
Makefile.am: Do not remove ChangeLog on make distclean, fixes #116
[pbatard: Remove it on make maintainer-clean instead]
Vitali Lovich [Wed, 14 Sep 2011 05:02:06 +0000 (07:02 +0200)]
Darwin: Reduce race likelihood between cancellation and device disconnect
References #88. The race condition still remains, but this change
makes it less likely to trigger.
Peter Stuge [Wed, 14 Sep 2011 02:00:33 +0000 (04:00 +0200)]
Document that libusb_get_device_list() can return any libusb_error code
Fixes #87.
Pete Batard [Thu, 16 Jun 2011 09:49:13 +0000 (10:49 +0100)]
Linux: Fix #81 URB double free race condition on device disconnect
A submitted transfer that has just been failed by the kernel could be
picked up by an event handler to be cleaned up, where freeing of URB
memory would race with the submit function doing it's own cleanup and
freeing as a result of the submit failing.
libusb_submit_transfer() always holds itransfer->lock, so the race can
be avoided by taking that lock also in the cleanup path and checking
that the URB memory has not already been freed before freeing it there.
As http://libusb.org/ticket/81#comment:14 notes there is still another
possible, but unlikely, race condition between libusb_submit_transfer()
and an event handling thread. That will require more work to solve.
[stuge: Add check in cleanup path that URBs have not already been freed]
Peter Stuge [Wed, 14 Sep 2011 01:00:14 +0000 (03:00 +0200)]
Add libusb_error_name() API function
Until we have i18ned error messages we can at least offer easy
translation from numeric error code to the error code's name.
[hansg: switch() on enum to get a compiler warning if a case is missing]
[hansg: Better return value when called with an unknown error code]
[hansg: Fix DEFAULT_VISIBILITY in definition to compile on all platforms]
Peter Stuge [Wed, 14 Sep 2011 00:30:18 +0000 (02:30 +0200)]
Add libusb_has_capability() API function
Since it is currently not planned to change the filename of the libusb
library between libusb-1.0 versions at least on Windows systems it's
important to have a capability API.
Currently there exists only one capability; LIBUSB_CAP_HAS_CAPABILITY,
which tests support for the libusb_has_capability() API.
Sebastian Pipping [Tue, 13 Sep 2011 18:38:26 +0000 (20:38 +0200)]
Fix typos in introductory documentation
Pekka Nikander [Tue, 13 Sep 2011 16:57:43 +0000 (18:57 +0200)]
Linux: Add libusb_open() debug and error messages with the device path
Fixes #109.
Pekka Nikander [Tue, 13 Sep 2011 16:52:36 +0000 (18:52 +0200)]
Add debug message with libusb_open() error code
References #109.
Hans de Goede [Mon, 12 Sep 2011 16:11:20 +0000 (18:11 +0200)]
Document libusb_handle_events_completed() and _timeout_completed()
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
[stuge: Note that the old racy functions should be avoided by new code]
Graeme Gill [Sat, 10 Sep 2011 13:47:05 +0000 (15:47 +0200)]
Fix #56 race condition causing delayed completion of sync transfers
The sync API had a race where it would check a condition to know if it
needed to call a libusb_handle_events() function. However, the check
was done outside of the lock that is held while the condition is set,
so another thread could completely serve whatever was needed to make
the condition true between it being checked and the event handler being
called. This situation would be detected after a libusb-internal timeout
of 60 seconds, after which the transfer would be completed without
error, but with significant delay.
Original patch at http://marc.info/?l=libusb-devel&m=
127252114815709
Changes by Hans de Goede:
- Renamed the "race-proof" functions from libusb_handle_events*_check()
to libusb_handle_events*_completed()
- Drop r = 0 setting in libusb_handle_events_timeout_completed()
(to make both completed checking cases identical flow wise)
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
[stuge: Simplify libusb_handle_events_timeout() change with a goto]
[pbatard: Fix _handle_events_timeout() and _completed() definitions]