[dpdk-dev] [PATCH v4] eal/windows: fix thread handle

Tasnim Bashar tbashar at mellanox.com
Thu Jun 25 21:11:54 CEST 2020


> From: Ranjit Menon <ranjit.menon at intel.com>
> Sent: Wednesday, June 24, 2020 5:39 PM
> To: Tasnim Bashar <tbashar at mellanox.com>; dev at dpdk.org
> Cc: harini.ramakrishnan at microsoft.com; pallavi.kadam at intel.com;
> ocardona at microsoft.com; navasile at linux.microsoft.com;
> dmitry.kozliuk at gmail.com; Tal Shnaiderman <talshn at mellanox.com>; Fady
> Bader <fady at mellanox.com>; Ophir Munk <ophirmu at mellanox.com>; Thomas
> Monjalon <thomas at monjalon.net>
> Subject: Re: [PATCH v4] eal/windows: fix thread handle
> 
> Hi, Tasnim...
> 
> On 6/24/2020 4:10 PM, Tasnim Bashar wrote:
> > Casting thread ID to handle is not accurate way to get thread handle.
> > Need to use OpenThread function to get thread handle from thread ID.
> >
> > pthread_setaffinity_np and pthread_getaffinity_np functions for
> > Windows are affected because of it.
> >
> > Signed-off-by: Tasnim Bashar <tbashar at mellanox.com>
> > ---
> > v3: WA to remove warning(-Wmaybe-uninitialized)
> > v4: Directly used function name instead of #define
> > ---
> >   lib/librte_eal/windows/include/pthread.h     | 84 +++++++++++++++-----
> >   lib/librte_eal/windows/include/rte_windows.h |  1 +
> >   2 files changed, 67 insertions(+), 18 deletions(-)
> >
> > diff --git a/lib/librte_eal/windows/include/pthread.h
> > b/lib/librte_eal/windows/include/pthread.h
> > index e2274cf4e9..2553c0890a 100644
> > --- a/lib/librte_eal/windows/include/pthread.h
> > +++ b/lib/librte_eal/windows/include/pthread.h
> > @@ -6,6 +6,7 @@
> >   #define _PTHREAD_H_
> >
> >   #include <stdint.h>
> > +#include <sched.h>
> >
> >   /**
> >    * This file is required to support the common code in
> > eal_common_proc.c, @@ -16,8 +17,8 @@
> >   extern "C" {
> >   #endif
> >
> > -#include <windows.h>
> >   #include <rte_common.h>
> > +#include <rte_windows.h>
> >
> >   #define PTHREAD_BARRIER_SERIAL_THREAD TRUE
> >
> > @@ -40,37 +41,84 @@ typedef SYNCHRONIZATION_BARRIER
> pthread_barrier_t;
> >   /* pthread function overrides */
> >   #define pthread_self() \
> >   	((pthread_t)GetCurrentThreadId())
> > -#define pthread_setaffinity_np(thread, size, cpuset) \
> > -	eal_set_thread_affinity_mask(thread, (unsigned long *) cpuset)
> > -#define pthread_getaffinity_np(thread, size, cpuset) \
> > -	eal_get_thread_affinity_mask(thread, (unsigned long *) cpuset)
> > -#define pthread_create(threadid, threadattr, threadfunc, args) \
> > -	eal_create_thread(threadid, threadattr, threadfunc, args)
> >
> >   static inline int
> > -eal_set_thread_affinity_mask(pthread_t threadid, unsigned long
> > *cpuset)
> > +pthread_setaffinity_np(pthread_t threadid, size_t cpuset_size,
> > +			rte_cpuset_t *cpuset)
> >   {
> > -	SetThreadAffinityMask((HANDLE) threadid, *cpuset);
> > -	return 0;
> > +	DWORD_PTR ret;
> > +	HANDLE thread_handle;
> > +
> > +	if (cpuset == NULL || cpuset_size == 0)
> > +		return -1;
> > +
> > +	thread_handle = OpenThread(THREAD_ALL_ACCESS, FALSE, threadid);
> > +	if (thread_handle == NULL) {
> > +		RTE_LOG_WIN32_ERR("OpenThread()");
> > +		return -1;
> > +	}
> > +
> > +	ret = SetThreadAffinityMask(thread_handle, *cpuset->_bits);
> > +	if (ret == 0) {
> > +		RTE_LOG_WIN32_ERR("SetThreadAffinityMask()");
> > +		goto close_handle;
> > +	}
> > +
> > +close_handle:
> > +	if (CloseHandle(thread_handle) == 0) {
> > +		RTE_LOG_WIN32_ERR("CloseHandle()");
> > +		return -1;
> > +	}
> > +	return (ret == 0) ? -1 : 0;
> >   }
> >
> >   static inline int
> > -eal_get_thread_affinity_mask(pthread_t threadid, unsigned long
> > *cpuset)
> > +pthread_getaffinity_np(pthread_t threadid, size_t cpuset_size,
> > +			rte_cpuset_t *cpuset)
> >   {
> >   	/* Workaround for the lack of a GetThreadAffinityMask()
> >   	 *API in Windows
> >   	 */
> > -		/* obtain previous mask by setting dummy mask */
> > -	DWORD dwprevaffinitymask =
> > -		SetThreadAffinityMask((HANDLE) threadid, 0x1);
> > +	DWORD_PTR prev_affinity_mask;
> > +	HANDLE thread_handle;
> > +	DWORD_PTR ret;
> Initialize ret to 0 here, otherwise... (see below)
> > +
> > +	if (cpuset == NULL || cpuset_size == 0)
> > +		return -1;
> > +
> > +	thread_handle = OpenThread(THREAD_ALL_ACCESS, FALSE, threadid);
> > +	if (thread_handle == NULL) {
> > +		RTE_LOG_WIN32_ERR("OpenThread()");
> > +		return -1;
> > +	}
> > +
> > +	/* obtain previous mask by setting dummy mask */
> > +	prev_affinity_mask = SetThreadAffinityMask(thread_handle, 0x1);
> > +	if (prev_affinity_mask == 0) {
> > +		RTE_LOG_WIN32_ERR("SetThreadAffinityMask()");
> > +		goto close_handle;
> ...after this jump, ret is uninitialized and will produce random results at the check
> at the end of the function.

Good observation Ranjit. I will send new patch with the fix.
> > +	}
> > +
> >   	/* set it back! */
> > -	SetThreadAffinityMask((HANDLE) threadid, dwprevaffinitymask);
> > -	*cpuset = dwprevaffinitymask;
> > -	return 0;
> > +	ret = SetThreadAffinityMask(thread_handle, prev_affinity_mask);
> > +	if (ret == 0) {
> > +		RTE_LOG_WIN32_ERR("SetThreadAffinityMask()");
> > +		goto close_handle;
> > +	}
> > +
> > +	memset(cpuset, 0, cpuset_size);
> > +	*cpuset->_bits = prev_affinity_mask;
> > +
> > +close_handle:
> > +	if (CloseHandle(thread_handle) == 0) {
> > +		RTE_LOG_WIN32_ERR("SetThreadAffinityMask()");
> > +		return -1;
> > +	}
> > +	return (ret == 0) ? -1 : 0;
> >   }
> >
> >   static inline int
> > -eal_create_thread(void *threadid, const void *threadattr, void
> > *threadfunc,
> > +pthread_create(void *threadid, const void *threadattr, void
> > +*threadfunc,
> >   		void *args)
> >   {
> >   	RTE_SET_USED(threadattr);
> > diff --git a/lib/librte_eal/windows/include/rte_windows.h
> > b/lib/librte_eal/windows/include/rte_windows.h
> > index 899ed7d874..d013b50241 100644
> > --- a/lib/librte_eal/windows/include/rte_windows.h
> > +++ b/lib/librte_eal/windows/include/rte_windows.h
> > @@ -31,6 +31,7 @@
> >   #define INITGUID
> >   #endif
> >   #include <initguid.h>
> > +#include <rte_log.h>
> >
> >   /**
> >    * Log GetLastError() with context, usually a Win32 API function and
> arguments.
> 
> 
> ranjit m.



More information about the dev mailing list