Re: [RFC][PATCH] io_uring: Incapsulate conditional mem accounting

[Pavel Begunkov]

ping

On 11/07/2019 04:07, Pavel Begunkov (Silence) wrote:
> From: Pavel Begunkov <asml.silence@xxxxxxxxx>
> 
> It's quite tedious and error-prone to manually check before each call
> to io_{,un}account_mem() whether we need memory accounting. Instead,
> the functions can work directly with struct io_ring_ctx and handle
> checks themselves. In any case, they're perfectly inlined.
> 
> Signed-off-by: Pavel Begunkov <asml.silence@xxxxxxxxx>
> ---
>  fs/io_uring.c | 59 ++++++++++++++++++++++-----------------------------
>  1 file changed, 25 insertions(+), 34 deletions(-)
> 
> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index 3fd884b4e0be..f47f7abe19eb 100644
> --- a/fs/io_uring.c
> +++ b/fs/io_uring.c
> @@ -2704,14 +2704,19 @@ static int io_sq_offload_start(struct io_ring_ctx *ctx,
>  	return ret;
>  }
>  
> -static void io_unaccount_mem(struct user_struct *user, unsigned long nr_pages)
> +static void io_unaccount_mem(struct io_ring_ctx *ctx, unsigned long nr_pages)
>  {
> -	atomic_long_sub(nr_pages, &user->locked_vm);
> +	if (ctx->account_mem)
> +		atomic_long_sub(nr_pages, &ctx->user->locked_vm);
>  }
>  
> -static int io_account_mem(struct user_struct *user, unsigned long nr_pages)
> +static int io_account_mem(struct io_ring_ctx *ctx, unsigned long nr_pages)
>  {
>  	unsigned long page_limit, cur_pages, new_pages;
> +	struct user_struct *user = ctx->user;
> +
> +	if (!ctx->account_mem)
> +		return 0;
>  
>  	/* Don't allow more pages than we can safely lock */
>  	page_limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT;
> @@ -2773,8 +2778,7 @@ static int io_sqe_buffer_unregister(struct io_ring_ctx *ctx)
>  		for (j = 0; j < imu->nr_bvecs; j++)
>  			put_page(imu->bvec[j].bv_page);
>  
> -		if (ctx->account_mem)
> -			io_unaccount_mem(ctx->user, imu->nr_bvecs);
> +		io_unaccount_mem(ctx, imu->nr_bvecs);
>  		kvfree(imu->bvec);
>  		imu->nr_bvecs = 0;
>  	}
> @@ -2857,11 +2861,9 @@ static int io_sqe_buffer_register(struct io_ring_ctx *ctx, void __user *arg,
>  		start = ubuf >> PAGE_SHIFT;
>  		nr_pages = end - start;
>  
> -		if (ctx->account_mem) {
> -			ret = io_account_mem(ctx->user, nr_pages);
> -			if (ret)
> -				goto err;
> -		}
> +		ret = io_account_mem(ctx, nr_pages);
> +		if (ret)
> +			goto err;
>  
>  		ret = 0;
>  		if (!pages || nr_pages > got_pages) {
> @@ -2874,8 +2876,7 @@ static int io_sqe_buffer_register(struct io_ring_ctx *ctx, void __user *arg,
>  					GFP_KERNEL);
>  			if (!pages || !vmas) {
>  				ret = -ENOMEM;
> -				if (ctx->account_mem)
> -					io_unaccount_mem(ctx->user, nr_pages);
> +				io_unaccount_mem(ctx, nr_pages);
>  				goto err;
>  			}
>  			got_pages = nr_pages;
> @@ -2885,8 +2886,7 @@ static int io_sqe_buffer_register(struct io_ring_ctx *ctx, void __user *arg,
>  						GFP_KERNEL);
>  		ret = -ENOMEM;
>  		if (!imu->bvec) {
> -			if (ctx->account_mem)
> -				io_unaccount_mem(ctx->user, nr_pages);
> +			io_unaccount_mem(ctx, nr_pages);
>  			goto err;
>  		}
>  
> @@ -2919,8 +2919,7 @@ static int io_sqe_buffer_register(struct io_ring_ctx *ctx, void __user *arg,
>  				for (j = 0; j < pret; j++)
>  					put_page(pages[j]);
>  			}
> -			if (ctx->account_mem)
> -				io_unaccount_mem(ctx->user, nr_pages);
> +			io_unaccount_mem(ctx, nr_pages);
>  			kvfree(imu->bvec);
>  			goto err;
>  		}
> @@ -3009,9 +3008,7 @@ static void io_ring_ctx_free(struct io_ring_ctx *ctx)
>  	io_mem_free(ctx->cq_ring);
>  
>  	percpu_ref_exit(&ctx->refs);
> -	if (ctx->account_mem)
> -		io_unaccount_mem(ctx->user,
> -				ring_pages(ctx->sq_entries, ctx->cq_entries));
> +	io_unaccount_mem(ctx, ring_pages(ctx->sq_entries, ctx->cq_entries));
>  	free_uid(ctx->user);
>  	kfree(ctx);
>  }
> @@ -3253,7 +3250,6 @@ static int io_uring_create(unsigned entries, struct io_uring_params *p)
>  {
>  	struct user_struct *user = NULL;
>  	struct io_ring_ctx *ctx;
> -	bool account_mem;
>  	int ret;
>  
>  	if (!entries || entries > IORING_MAX_ENTRIES)
> @@ -3269,29 +3265,24 @@ static int io_uring_create(unsigned entries, struct io_uring_params *p)
>  	p->cq_entries = 2 * p->sq_entries;
>  
>  	user = get_uid(current_user());
> -	account_mem = !capable(CAP_IPC_LOCK);
> -
> -	if (account_mem) {
> -		ret = io_account_mem(user,
> -				ring_pages(p->sq_entries, p->cq_entries));
> -		if (ret) {
> -			free_uid(user);
> -			return ret;
> -		}
> -	}
>  
>  	ctx = io_ring_ctx_alloc(p);
>  	if (!ctx) {
> -		if (account_mem)
> -			io_unaccount_mem(user, ring_pages(p->sq_entries,
> -								p->cq_entries));
>  		free_uid(user);
>  		return -ENOMEM;
>  	}
> +
>  	ctx->compat = in_compat_syscall();
> -	ctx->account_mem = account_mem;
> +	ctx->account_mem = !capable(CAP_IPC_LOCK);
>  	ctx->user = user;
>  
> +	ret = io_account_mem(ctx, ring_pages(p->sq_entries, p->cq_entries));
> +	if (ret) {
> +		free_uid(user);
> +		kfree(ctx);
> +		return ret;
> +	}
> +
>  	ret = io_allocate_scq_urings(ctx, p);
>  	if (ret)
>  		goto err;
> 

-- 
Yours sincerely,
Pavel Begunkov

Attachment: signature.asc
Description: OpenPGP digital signature