| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <56A92246.7010807@gmail.com>
Date: Wed, 27 Jan 2016 12:02:14 -0800
From: Florian Fainelli <f.fainelli@...il.com>
To: Gregory CLEMENT <gregory.clement@...e-electrons.com>,
"David S. Miller" <davem@...emloft.net>,
linux-kernel@...r.kernel.org, netdev@...r.kernel.org,
Thomas Petazzoni <thomas.petazzoni@...e-electrons.com>
CC: Jason Cooper <jason@...edaemon.net>, Andrew Lunn <andrew@...n.ch>,
Sebastian Hesselbarth <sebastian.hesselbarth@...il.com>,
linux-arm-kernel@...ts.infradead.org,
Lior Amsalem <alior@...vell.com>,
Nadav Haklai <nadavh@...vell.com>,
Marcin Wojtas <mw@...ihalf.com>,
Simon Guinot <simon.guinot@...uanux.org>,
Ezequiel Garcia <ezequiel.garcia@...e-electrons.com>,
Maxime Ripard <maxime.ripard@...e-electrons.com>,
Boris BREZILLON <boris.brezillon@...e-electrons.com>,
Russell King - ARM Linux <linux@....linux.org.uk>,
Willy Tarreau <w@....eu>, Arnd Bergmann <arnd@...db.de>
Subject: Re: [PATCH net-next 09/10] net: Add a hardware buffer management
helper API
On 12/01/16 11:10, Gregory CLEMENT wrote:
> This basic implementation allows to share code between driver using
> hardware buffer management. As the code is hardware agnostic, there is
> few helpers, most of the optimization brought by the an HW BM has to be
> done at driver level.
>
> Signed-off-by: Gregory CLEMENT <gregory.clement@...e-electrons.com>
> ---
> include/net/hwbm.h | 19 +++++++++++++
> net/core/Makefile | 2 +-
> net/core/hwbm.c | 78 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 3 files changed, 98 insertions(+), 1 deletion(-)
> create mode 100644 include/net/hwbm.h
> create mode 100644 net/core/hwbm.c
>
> diff --git a/include/net/hwbm.h b/include/net/hwbm.h
> new file mode 100644
> index 000000000000..898ccd2fb58d
> --- /dev/null
> +++ b/include/net/hwbm.h
> @@ -0,0 +1,19 @@
> +#ifndef _HWBM_H
> +#define _HWBM_H
> +
> +struct hwbm_pool {
> + /* Size of the buffers managed */
> + int size;
> + /* Number of buffers currently used by this pool */
> + int buf_num;
> + /* constructor called during alocation */
> + int (*construct)(struct hwbm_pool *bm_pool, void *buf);
Having the buffer size might be handy too.
> + /* private data */
> + void *priv;
> +};
> +
> +void hwbm_buf_free(struct hwbm_pool *bm_pool, void *buf);
> +int hwbm_pool_refill(struct hwbm_pool *bm_pool);
> +int hwbm_pool_add(struct hwbm_pool *bm_pool, int buf_num);
> +
> +#endif /* _HWBM_H */
> diff --git a/net/core/Makefile b/net/core/Makefile
> index 0b835de04de3..df81bf11f072 100644
> --- a/net/core/Makefile
> +++ b/net/core/Makefile
> @@ -9,7 +9,7 @@ obj-$(CONFIG_SYSCTL) += sysctl_net_core.o
>
> obj-y += dev.o ethtool.o dev_addr_lists.o dst.o netevent.o \
> neighbour.o rtnetlink.o utils.o link_watch.o filter.o \
> - sock_diag.o dev_ioctl.o tso.o sock_reuseport.o
> + sock_diag.o dev_ioctl.o tso.o sock_reuseport.o hwbm.o
Not everybody will want this built in by default, we probably need a
hidden config symbol here.
>
> obj-$(CONFIG_XFRM) += flow.o
> obj-y += net-sysfs.o
> diff --git a/net/core/hwbm.c b/net/core/hwbm.c
> new file mode 100644
> index 000000000000..d5d40d63cb34
> --- /dev/null
> +++ b/net/core/hwbm.c
> @@ -0,0 +1,78 @@
> +/* Support for hardware buffer manager.
> + *
> + * Copyright (C) 2016 Marvell
> + *
> + * Gregory CLEMENT <gregory.clement@...e-electrons.com>
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> + */
> +#include <linux/kernel.h>
> +#include <linux/printk.h>
> +#include <linux/skbuff.h>
> +#include <net/hwbm.h>
> +
> +void hwbm_buf_free(struct hwbm_pool *bm_pool, void *buf)
> +{
> + if (likely(bm_pool->size <= PAGE_SIZE))
> + skb_free_frag(buf);
> + else
> + kfree(buf);
> +}
> +EXPORT_SYMBOL_GPL(hwbm_buf_free);
> +
> +/* Refill processing for HW buffer management */
> +int hwbm_pool_refill(struct hwbm_pool *bm_pool)
> +{
> + void *buf;
> + int frag_size = bm_pool->size;
Reverse christmas tree declaration looks a bit nicer.
> +
> + if (likely(frag_size <= PAGE_SIZE))
> + buf = netdev_alloc_frag(frag_size);
> + else
> + buf = kmalloc(frag_size, GFP_ATOMIC);
Maybe we should allow the caller to specify a gfp_t, just in case
GFP_ATOMIC is not good enough.
> +
> + if (!buf)
> + return -ENOMEM;
> +
> + if (bm_pool->construct)
> + if (bm_pool->construct(bm_pool, buf)) {
> + hwbm_buf_free(bm_pool, buf);
> + return -ENOMEM;
> + }
> +
> + return 0;
> +}
> +EXPORT_SYMBOL_GPL(hwbm_pool_refill);
> +
> +int hwbm_pool_add(struct hwbm_pool *bm_pool, int buf_num)
unsigned int buf_num
> +{
> + int err, i;
> +
> + if (bm_pool->buf_num == bm_pool->size) {
> + pr_debug("pool already filled\n");
> + return bm_pool->buf_num;
> + }
> +
> + if (buf_num + bm_pool->buf_num > bm_pool->size) {
> + pr_debug("cannot allocate %d buffers for pool\n",
> + buf_num);
> + return 0;
> + }
buf_num is under caller control, and potentially hardware control
indirectly, what if I make this arbitrary big and wrap around?
> +
> + for (i = 0; i < buf_num; i++) {
> + err = hwbm_pool_refill(bm_pool);
> + if (err < 0)
> + break;
> + }
If we fail refiling here, should not we propagate the error back to the
caller?
> +
> + /* Update BM driver with number of buffers added to pool */
> + bm_pool->buf_num += i;
> +
> + pr_debug("hwpm pool: %d of %d buffers added\n", i, buf_num);
No locking or atomic operations here? What if two CPUs call into this
function?
> +
> + return i;
> +}
> +EXPORT_SYMBOL_GPL(hwbm_pool_add);
>
--
Florian
Powered by blists - more mailing lists