From: Ben Peart <peartben@gmail.com>
Date: Mon, 29 Oct 2018 16:52:56 GMT
Subject: Re: [PATCH 07/10] preload-index.c: remove #ifdef NO_PTHREADS
Message-ID: <1217c887-cb15-87c6-649f-8b927abe8180@gmail.com>
In-Reply-To: <20181027071003.1347-8-pclouds@gmail.com>



On 10/27/2018 3:10 AM, Nguyễn Thái Ngọc Duy wrote:
> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
> ---
>   preload-index.c | 15 ++-------------
>   1 file changed, 2 insertions(+), 13 deletions(-)
> 
> diff --git a/preload-index.c b/preload-index.c
> index 9e7152ab14..0e24886aca 100644
> --- a/preload-index.c
> +++ b/preload-index.c
> @@ -7,17 +7,7 @@
>   #include "fsmonitor.h"
>   #include "config.h"
>   #include "progress.h"
> -
> -#ifdef NO_PTHREADS
> -static void preload_index(struct index_state *index,
> -			  const struct pathspec *pathspec,
> -			  unsigned int refresh_flags)
> -{
> -	; /* nothing */
> -}
> -#else
> -
> -#include <pthread.h>
> +#include "thread-utils.h"
>   
>   /*
>    * Mostly randomly chosen maximum thread counts: we
> @@ -108,7 +98,7 @@ static void preload_index(struct index_state *index,
>   	struct thread_data data[MAX_PARALLEL];
>   	struct progress_data pd;
>   
> -	if (!core_preload_index)
> +	if (!HAVE_THREADS || !core_preload_index)

I also "hoped in general that the non-threaded code paths could mostly 
just collapse into the same as the "threads == 1" cases, and we wouldn't 
have to "are threads even supported" in a lot of places."

In this case, if we cap the threads to online_cpus() the later test for 
'if (threads < 2)' would do that.  I haven't measured this specific case 
but in the other cases I have measured, having more threads than cpus 
did not result in a performance win.

>   		return;
>   
>   	threads = index->cache_nr / THREAD_COST;
> @@ -151,7 +141,6 @@ static void preload_index(struct index_state *index,
>   
>   	trace_performance_leave("preload index");
>   }
> -#endif
>   
>   int read_index_preload(struct index_state *index,
>   		       const struct pathspec *pathspec,
>