[PATCH] eal/windows: fix data race when creating threads
Dmitry Kozlyuk
dmitry.kozliuk at gmail.com
Thu Mar 10 17:19:32 CET 2022
Hi Tyler,
2022-03-09 23:35 (UTC-0800), Tyler Retzlaff:
> create lcore worker threads suspended and then subsequently resume to
> allow &lcore_config[i].thread_id be stored before eal_thread_loop
> execution.
>
> Fixes: 53ffd9f080fc ("eal/windows: add minimum viable code")
> Cc: anand.rawat at intel.com
> Cc: stable at dpdk.org
>
> Signed-off-by: Tyler Retzlaff <roretzla at linux.microsoft.com>
Acked-by: Dmitry Kozlyuk <dmitry.kozliuk at gmail.com>
It takes some effort to correlate &lcore_config[i].thread_id
with the code in the patch and knowledge of CreateThread()
to understand the root cause.
I suggest prepending an explanation to your commit message:
eal_thread_loop() uses lcore_config[i].thread_id,
which is stored upon the return from CreateThread().
Per documentation, eal_thread_loop() can start
before CreateThread() returns and the ID is stored.
> ---
> lib/eal/windows/eal_thread.c | 7 ++++++-
> 1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/lib/eal/windows/eal_thread.c b/lib/eal/windows/eal_thread.c
> index 54fa93f..ff84cb4 100644
> --- a/lib/eal/windows/eal_thread.c
> +++ b/lib/eal/windows/eal_thread.c
> @@ -150,13 +150,18 @@
>
> th = CreateThread(NULL, 0,
> (LPTHREAD_START_ROUTINE)(ULONG_PTR)eal_thread_loop,
> - NULL, 0, (LPDWORD)thread);
> + NULL, CREATE_SUSPENDED, (LPDWORD)thread);
> if (!th)
> return -1;
>
> SetPriorityClass(GetCurrentProcess(), NORMAL_PRIORITY_CLASS);
> SetThreadPriority(th, THREAD_PRIORITY_NORMAL);
>
> + if (ResumeThread(th) == (DWORD)-1) {
> + (void)CloseHandle(th);
> + return -1;
> + }
> +
> return 0;
> }
>
More information about the stable
mailing list