Re: [PATCH v7 1/2]: perf util: map data buffer for preserving collected data

[Jiri Olsa]

On Wed, Sep 05, 2018 at 10:19:56AM +0300, Alexey Budankov wrote:

SNIP

> @@ -166,6 +174,7 @@ void perf_mmap__munmap(struct perf_mmap *map)
>  
>  int perf_mmap__mmap(struct perf_mmap *map, struct mmap_params *mp, int fd)
>  {
> +	int i;
>  	/*
>  	 * The last one will be done at perf_mmap__consume(), so that we
>  	 * make sure we don't prevent tools from consuming every last event in
> @@ -190,6 +199,50 @@ int perf_mmap__mmap(struct perf_mmap *map, struct mmap_params *mp, int fd)
>  		map->base = NULL;
>  		return -1;
>  	}
> +	map->nr_cblocks = mp->nr_cblocks;
> +	map->cblocks = calloc(map->nr_cblocks, sizeof(struct aiocb*));
> +	if (!map->cblocks) {
> +		pr_debug2("failed to allocate perf event data buffers, error %d\n",
> +				errno);
> +		return -1;
> +	}
> +	map->data = calloc(map->nr_cblocks, sizeof(void*));
> +	if (map->data) {
> +		int delta_max = sysconf(_SC_AIO_PRIO_DELTA_MAX);
> +		for (i = 0; i < map->nr_cblocks; ++i) {
> +			map->data[i] = malloc(perf_mmap__mmap_len(map));
> +			if (map->data[i]) {
> +				int prio;
> +				unsigned char *data = map->data[i];
> +				map->cblocks[i] = (struct aiocb *)&data[map->mask + 1];

the 'struct aiocb' is allocated at the last page of the buffer?
is that enough space? could we please make this more transparent
and allocate that space separately?

thanks,
jirka