* Re: [PATCH bpf-next] bpf, capabilities: introduce CAP_BPF
From: kbuild test robot @ 2019-08-28 10:38 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: kbuild-all, luto, davem, daniel, netdev, bpf, kernel-team,
linux-api
In-Reply-To: <20190827205213.456318-1-ast@kernel.org>
[-- Attachment #1: Type: text/plain, Size: 1553 bytes --]
Hi Alexei,
I love your patch! Yet something to improve:
[auto build test ERROR on bpf-next/master]
url: https://github.com/0day-ci/linux/commits/Alexei-Starovoitov/bpf-capabilities-introduce-CAP_BPF/20190828-142441
base: https://kernel.googlesource.com/pub/scm/linux/kernel/git/bpf/bpf-next.git master
config: ia64-defconfig (attached as .config)
compiler: ia64-linux-gcc (GCC) 7.4.0
reproduce:
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
GCC_VERSION=7.4.0 make.cross ARCH=ia64
If you fix the issue, kindly add following tag
Reported-by: kbuild test robot <lkp@intel.com>
All errors (new ones prefixed by >>):
kernel/bpf/core.c: In function 'cap_bpf_tracing':
>> kernel/bpf/core.c:2110:31: error: implicit declaration of function 'perf_paranoid_tracepoint_raw' [-Werror=implicit-function-declaration]
(capable(CAP_BPF) && !perf_paranoid_tracepoint_raw());
^~~~~~~~~~~~~~~~~~~~~~~~~~~~
cc1: some warnings being treated as errors
vim +/perf_paranoid_tracepoint_raw +2110 kernel/bpf/core.c
2106
2107 bool cap_bpf_tracing(void)
2108 {
2109 return capable(CAP_SYS_ADMIN) ||
> 2110 (capable(CAP_BPF) && !perf_paranoid_tracepoint_raw());
2111 }
2112
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 19214 bytes --]
^ permalink raw reply
* RE: Query on possible bug in the can_create_echo_skb() API
From: Srinivas Neeli @ 2019-08-28 11:02 UTC (permalink / raw)
To: Marc Kleine-Budde, wg@grandegger.com
Cc: Srinivas Goud, Naga Sureshkumar Relli,
Appana Durga Kedareswara Rao, linux-can@vger.kernel.org,
netdev@vger.kernel.org, linux-kernel@vger.kernel.org
In-Reply-To: <6bd3a657-dd8a-03a5-1e7c-bac532008f6e@pengutronix.de>
Hi,
Case 1:
can_put_echo_skb(); -> skb = can_create_echo_skb(skb); -> return skb;
In can_create_echo_skb() not using the shared_skb, so we are returning the old skb.
Storing the return value in "skb". But it's a pointer, for storing that need double pointer.
Instead of double-pointer using a single pointer. In this scenario it's ok , we are returning the same SKB.
Case 2:
can_put_echo_skb(skb, ndev, priv->tx_head % priv->tx_max); -> skb = can_create_echo_skb(skb); -> can_skb_set_owner(nskb, skb->sk); - Returning nskb;
shared_skb scenario:
In share-skb case “can_create_echo_skb(skb);” returning "new skb". For storing new skb need a double pointer.
Providing an example for overcoming above issue.
Example:
can_put_echo_skb(struct sk_buff **skb,struct net_device *dev,unsigned int idx);
If you ok with this change, I will send a patch.
Thanks
Srinivas Neeli
> -----Original Message-----
> From: Marc Kleine-Budde <mkl@pengutronix.de>
> Sent: Wednesday, August 28, 2019 1:03 AM
> To: Srinivas Neeli <sneeli@xilinx.com>; wg@grandegger.com
> Cc: Srinivas Goud <sgoud@xilinx.com>; Naga Sureshkumar Relli
> <nagasure@xilinx.com>; Appana Durga Kedareswara Rao
> <appanad@xilinx.com>; linux-can@vger.kernel.org
> Subject: Re: Query on possible bug in the can_create_echo_skb() API
>
> Hello Srinivas Neeli,
>
> please don't send HTML messages to the kernel mailinglists.
>
> On 8/21/19 12:51 PM, Srinivas Neeli wrote:
> > While walking through the CAN core layer dev.c file in the
> > can_put_echo_skb() API [1], Seems to be there is a race condition in
> > the
> > can_create_echo_skb() API, more details below
> >
> > If the skb is a shared skb, we are overwriting the skb pointer [2] in
> > the can_create_echo_skb() API and returning the new skb back.
>
> Where and how is the skb pointer overwritten? Can you explain a bit more.
>
> > If the core layer/drivers use this skb it is not valid any more (it
> > may lead to crash/oops).
> >
> >
> >
> > A possible solution for this issue would make the function input
> > argument should be double-pointer.
> >
> > Please correct me if my analyzation is wrong.
>
> Can you provide a patch of your proposed changes?
>
> regards,
> Marc
>
> --
> Pengutronix e.K. | Marc Kleine-Budde |
> Industrial Linux Solutions | Phone: +49-231-2826-924 |
> Vertretung West/Dortmund | Fax: +49-5121-206917-5555 |
> Amtsgericht Hildesheim, HRA 2686 | http://www.pengutronix.de |
^ permalink raw reply
* multipath tcp MIB counter placement - share with tcp or extra?
From: Florian Westphal @ 2019-08-28 11:43 UTC (permalink / raw)
To: Eric Dumazet; +Cc: netdev
Hi Eric,
The out-of-tree multipath TCP stack adds a few MIB counters to track
(and debug) MTPCP behaviour. Examples:
SNMP_MIB_ITEM("MPCapableSYNRX", MPTCP_MIB_MPCAPABLEPASSIVE),
SNMP_MIB_ITEM("MPCapableSYNTX", MPTCP_MIB_MPCAPABLEACTIVE),
[..]
SNMP_MIB_ITEM("MPTCPRetrans", MPTCP_MIB_RETRANSSEGS),
SNMP_MIB_ITEM("MPFailRX", MPTCP_MIB_MPFAILRX),
SNMP_MIB_ITEM("MPCsumFail", MPTCP_MIB_CSUMFAIL),
and so on.
I think that such MIB counters would be good to have in the 'upstreaming'
attempt as well.
The out-of-tree code keeps them separate from the tcp mib counters and also
exposes them in a different /proc file (/proc/net/mptcp_net/snmp).
Would you be ok with mptcp-upstreaming adding its MIB counters to the
existing TCP MIB instead?
This would make 'nstat' and other tools pick them up automatically.
It would also help TCP highlevel debugging to see if MPTCP is involved
in any way.
Let me know -- I can go with a separate MIB, its no problem, I just want
to avoid going down the wrong path.
Thanks,
Florian
^ permalink raw reply
* Re: [PATCH net-next v2 3/3] dpaa2-eth: Add pause frame support
From: Andrew Lunn @ 2019-08-28 11:52 UTC (permalink / raw)
To: Ioana Ciocoi Radulescu
Cc: netdev@vger.kernel.org, davem@davemloft.net, Ioana Ciornei
In-Reply-To: <AM0PR04MB499496AC09FD7BE58AE7B9C394A30@AM0PR04MB4994.eurprd04.prod.outlook.com>
> Clearing the ASYM_PAUSE flag only means we tell the firmware we want
> both Rx and Tx pause to be enabled in the beginning. User can still set
> an asymmetric config (i.e. only Rx pause or only Tx pause to be enabled)
> if needed.
>
> The truth table is like this:
>
> PAUSE | ASYM_PAUSE | Rx pause | Tx pause
> ----------------------------------------
> 0 | 0 | disabled | disabled
> 0 | 1 | disabled | enabled
> 1 | 0 | enabled | enabled
> 1 | 1 | enabled | disabled
Hi Ioana
Ah, that is not intuitive. Please add a comment, and maybe this table
to the commit message.
Thanks
Andrew
^ permalink raw reply
* Re: [PATCH net-next v3 05/10] net: sched: add API for registering unlocked offload block callbacks
From: tanhuazhong @ 2019-08-28 11:53 UTC (permalink / raw)
To: Vlad Buslov, netdev
Cc: jhs, xiyou.wangcong, jiri, davem, jakub.kicinski, pablo,
Jiri Pirko
In-Reply-To: <20190826134506.9705-6-vladbu@mellanox.com>
On 2019/8/26 21:45, Vlad Buslov wrote:
> Extend struct flow_block_offload with "unlocked_driver_cb" flag to allow
> registering and unregistering block hardware offload callbacks that do not
> require caller to hold rtnl lock. Extend tcf_block with additional
> lockeddevcnt counter that is incremented for each non-unlocked driver
> callback attached to device. This counter is necessary to conditionally
> obtain rtnl lock before calling hardware callbacks in following patches.
>
> Register mlx5 tc block offload callbacks as "unlocked".
>
> Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
> Acked-by: Jiri Pirko <jiri@mellanox.com>
> ---
> drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 2 ++
> drivers/net/ethernet/mellanox/mlx5/core/en_rep.c | 3 +++
> include/net/flow_offload.h | 1 +
> include/net/sch_generic.h | 1 +
> net/sched/cls_api.c | 6 ++++++
> 5 files changed, 13 insertions(+)
>
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
> index fa4bf2d4bcd4..8592b98d0e70 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
> @@ -3470,10 +3470,12 @@ static int mlx5e_setup_tc(struct net_device *dev, enum tc_setup_type type,
> void *type_data)
> {
> struct mlx5e_priv *priv = netdev_priv(dev);
> + struct flow_block_offload *f = type_data;
>
> switch (type) {
> #ifdef CONFIG_MLX5_ESWITCH
> case TC_SETUP_BLOCK:
> + f->unlocked_driver_cb = true;
> return flow_block_cb_setup_simple(type_data,
> &mlx5e_block_cb_list,
> mlx5e_setup_tc_block_cb,
Hi,
I have got below warning when compiling the latest net-next:
drivers/net/ethernet/mellanox//mlx5/core/en_main.c:3473:29: warning:
unused variable ‘f’ [-Wunused-variable]
struct flow_block_offload *f = type_data;
Could this variable be defined within "#ifdef CONFIG_MLX5_ESWITCH"?
BTW, it seems varible f has not been used in any place in addition to
assigning true to its member unlocked_driver_cb in "case
TC_SETUP_BLOCK:". Maybe I have miss something about it:).
Huazhong.
Thanks.
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
> index 3c0d36b2b91c..e7ac6233037d 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
> @@ -763,6 +763,7 @@ mlx5e_rep_indr_setup_tc_block(struct net_device *netdev,
> if (f->binder_type != FLOW_BLOCK_BINDER_TYPE_CLSACT_INGRESS)
> return -EOPNOTSUPP;
>
> + f->unlocked_driver_cb = true;
> f->driver_block_list = &mlx5e_block_cb_list;
>
> switch (f->command) {
> @@ -1245,9 +1246,11 @@ static int mlx5e_rep_setup_tc(struct net_device *dev, enum tc_setup_type type,
> void *type_data)
> {
> struct mlx5e_priv *priv = netdev_priv(dev);
> + struct flow_block_offload *f = type_data;
>
> switch (type) {
> case TC_SETUP_BLOCK:
> + f->unlocked_driver_cb = true;
> return flow_block_cb_setup_simple(type_data,
> &mlx5e_rep_block_cb_list,
> mlx5e_rep_setup_tc_cb,
> diff --git a/include/net/flow_offload.h b/include/net/flow_offload.h
> index 757fa84de654..fc881875f856 100644
> --- a/include/net/flow_offload.h
> +++ b/include/net/flow_offload.h
> @@ -284,6 +284,7 @@ struct flow_block_offload {
> enum flow_block_command command;
> enum flow_block_binder_type binder_type;
> bool block_shared;
> + bool unlocked_driver_cb;
> struct net *net;
> struct flow_block *block;
> struct list_head cb_list;
> diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
> index c4fbbaff30a2..43f5b7ed02bd 100644
> --- a/include/net/sch_generic.h
> +++ b/include/net/sch_generic.h
> @@ -408,6 +408,7 @@ struct tcf_block {
> bool keep_dst;
> atomic_t offloadcnt; /* Number of oddloaded filters */
> unsigned int nooffloaddevcnt; /* Number of devs unable to do offload */
> + unsigned int lockeddevcnt; /* Number of devs that require rtnl lock. */
> struct {
> struct tcf_chain *chain;
> struct list_head filter_chain_list;
> diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
> index 8b807e75fae2..1a39779bdbad 100644
> --- a/net/sched/cls_api.c
> +++ b/net/sched/cls_api.c
> @@ -1418,6 +1418,8 @@ static int tcf_block_bind(struct tcf_block *block,
> bo->extack);
> if (err)
> goto err_unroll;
> + if (!bo->unlocked_driver_cb)
> + block->lockeddevcnt++;
>
> i++;
> }
> @@ -1433,6 +1435,8 @@ static int tcf_block_bind(struct tcf_block *block,
> block_cb->cb_priv, false,
> tcf_block_offload_in_use(block),
> NULL);
> + if (!bo->unlocked_driver_cb)
> + block->lockeddevcnt--;
> }
> flow_block_cb_free(block_cb);
> }
> @@ -1454,6 +1458,8 @@ static void tcf_block_unbind(struct tcf_block *block,
> NULL);
> list_del(&block_cb->list);
> flow_block_cb_free(block_cb);
> + if (!bo->unlocked_driver_cb)
> + block->lockeddevcnt--;
> }
> }
>
>
^ permalink raw reply
* Re: Query on possible bug in the can_create_echo_skb() API
From: Marc Kleine-Budde @ 2019-08-28 12:15 UTC (permalink / raw)
To: Srinivas Neeli, wg@grandegger.com
Cc: Srinivas Goud, Naga Sureshkumar Relli,
Appana Durga Kedareswara Rao, linux-can@vger.kernel.org,
netdev@vger.kernel.org, linux-kernel@vger.kernel.org
In-Reply-To: <BYAPR02MB546403642B2233DDD5C456C4AFA30@BYAPR02MB5464.namprd02.prod.outlook.com>
[-- Attachment #1.1: Type: text/plain, Size: 1442 bytes --]
On 8/28/19 1:02 PM, Srinivas Neeli wrote:
> Case 1:
> can_put_echo_skb(); -> skb = can_create_echo_skb(skb); -> return skb;
>
> In can_create_echo_skb() not using the shared_skb, so we are returning the old skb.
> Storing the return value in "skb". But it's a pointer, for storing that need double pointer.
> Instead of double-pointer using a single pointer. In this scenario it's ok , we are returning the same SKB.
>
> Case 2:
> can_put_echo_skb(skb, ndev, priv->tx_head % priv->tx_max); -> skb = can_create_echo_skb(skb); -> can_skb_set_owner(nskb, skb->sk); - Returning nskb;
>
> shared_skb scenario:
> In share-skb case “can_create_echo_skb(skb);” returning "new skb". For storing new skb need a double pointer.
>
> Providing an example for overcoming above issue.
> Example:
> can_put_echo_skb(struct sk_buff **skb,struct net_device *dev,unsigned int idx);
Now I get what you mean.
> If you ok with this change, I will send a patch.
I think the can_put_echo_skb() API needs clarification. The driver is
not allowed to touch the skb any more after can_put_echo_skb(). We
should review the driver for this.
thanks,
Marc
--
Pengutronix e.K. | Marc Kleine-Budde |
Industrial Linux Solutions | Phone: +49-231-2826-924 |
Vertretung West/Dortmund | Fax: +49-5121-206917-5555 |
Amtsgericht Hildesheim, HRA 2686 | http://www.pengutronix.de |
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]
^ permalink raw reply
* Re: multipath tcp MIB counter placement - share with tcp or extra?
From: Eric Dumazet @ 2019-08-28 12:27 UTC (permalink / raw)
To: Florian Westphal, Eric Dumazet; +Cc: netdev
In-Reply-To: <20190828114321.GG20113@breakpoint.cc>
On 8/28/19 1:43 PM, Florian Westphal wrote:
> Hi Eric,
>
> The out-of-tree multipath TCP stack adds a few MIB counters to track
> (and debug) MTPCP behaviour. Examples:
>
> SNMP_MIB_ITEM("MPCapableSYNRX", MPTCP_MIB_MPCAPABLEPASSIVE),
> SNMP_MIB_ITEM("MPCapableSYNTX", MPTCP_MIB_MPCAPABLEACTIVE),
> [..]
> SNMP_MIB_ITEM("MPTCPRetrans", MPTCP_MIB_RETRANSSEGS),
> SNMP_MIB_ITEM("MPFailRX", MPTCP_MIB_MPFAILRX),
> SNMP_MIB_ITEM("MPCsumFail", MPTCP_MIB_CSUMFAIL),
>
> and so on.
>
> I think that such MIB counters would be good to have in the 'upstreaming'
> attempt as well.
>
> The out-of-tree code keeps them separate from the tcp mib counters and also
> exposes them in a different /proc file (/proc/net/mptcp_net/snmp).
>
> Would you be ok with mptcp-upstreaming adding its MIB counters to the
> existing TCP MIB instead?
>
> This would make 'nstat' and other tools pick them up automatically.
> It would also help TCP highlevel debugging to see if MPTCP is involved
> in any way.
>
> Let me know -- I can go with a separate MIB, its no problem, I just want
> to avoid going down the wrong path.
There are about 40 counters.
Space for that will be per netns : num_possible_cpus * 40 * 8 bytes
The cost of folding all the values will make nstat slower even if MPTCP is not used.
Maybe find a way to not having to fold the MPTCP percpu counters if MPTCP is not loaded ?
^ permalink raw reply
* Re: multipath tcp MIB counter placement - share with tcp or extra?
From: Florian Westphal @ 2019-08-28 12:32 UTC (permalink / raw)
To: Eric Dumazet; +Cc: Florian Westphal, Eric Dumazet, netdev
In-Reply-To: <deb00e41-0188-0ca9-ccb3-b74b34a4cc5d@gmail.com>
Eric Dumazet <eric.dumazet@gmail.com> wrote:
> > Let me know -- I can go with a separate MIB, its no problem, I just want
> > to avoid going down the wrong path.
>
> There are about 40 counters.
>
> Space for that will be per netns : num_possible_cpus * 40 * 8 bytes
>
> The cost of folding all the values will make nstat slower even if MPTCP is not used.
Ok, so 'same proc file' would be fine but 'increase pcpu mem cost
unconditionally' isn't.
> Maybe find a way to not having to fold the MPTCP percpu counters if MPTCP is not loaded ?
MPTCP is builtin (bool).
However, we might be able to delay allocation until first mptcp socket
is requested, I will see if this can be done somehow.
Thanks Eric!
^ permalink raw reply
* [PATCH net v4 1/2] Revert "r8152: napi hangup fix after disconnect"
From: Hayes Wang @ 2019-08-28 12:56 UTC (permalink / raw)
To: netdev; +Cc: nic_swsd, linux-kernel, Hayes Wang
In-Reply-To: <1394712342-15778-323-Taiwan-albertk@realtek.com>
This reverts commit 0ee1f4734967af8321ecebaf9c74221ace34f2d5.
The commit 0ee1f4734967 ("r8152: napi hangup fix after
disconnect") adds a check about RTL8152_UNPLUG to determine
if calling napi_disable() is invalid in rtl8152_close(),
when rtl8152_disconnect() is called. This avoids to use
napi_disable() after calling netif_napi_del().
However, commit ffa9fec30ca0 ("r8152: set RTL8152_UNPLUG
only for real disconnection") causes that RTL8152_UNPLUG
is not always set when calling rtl8152_disconnect().
Therefore, I have to revert commit 0ee1f4734967 ("r8152:
napi hangup fix after disconnect"), first. And submit
another patch to fix it.
Fixes: ffa9fec30ca0 ("r8152: set RTL8152_UNPLUG only for real disconnection")
Signed-off-by: Hayes Wang <hayeswang@realtek.com>
---
drivers/net/usb/r8152.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/drivers/net/usb/r8152.c b/drivers/net/usb/r8152.c
index eee0f5007ee3..ad3abe26b51b 100644
--- a/drivers/net/usb/r8152.c
+++ b/drivers/net/usb/r8152.c
@@ -4021,8 +4021,7 @@ static int rtl8152_close(struct net_device *netdev)
#ifdef CONFIG_PM_SLEEP
unregister_pm_notifier(&tp->pm_notifier);
#endif
- if (!test_bit(RTL8152_UNPLUG, &tp->flags))
- napi_disable(&tp->napi);
+ napi_disable(&tp->napi);
clear_bit(WORK_ENABLE, &tp->flags);
usb_kill_urb(tp->intr_urb);
cancel_delayed_work_sync(&tp->schedule);
--
2.21.0
^ permalink raw reply related
* [PATCH net v4 2/2] r8152: remove calling netif_napi_del
From: Hayes Wang @ 2019-08-28 12:56 UTC (permalink / raw)
To: netdev; +Cc: nic_swsd, linux-kernel, Hayes Wang
In-Reply-To: <1394712342-15778-323-Taiwan-albertk@realtek.com>
Remove unnecessary use of netif_napi_del. This also avoids to call
napi_disable() after netif_napi_del().
Fixes: ffa9fec30ca0 ("r8152: set RTL8152_UNPLUG only for real disconnection")
Signed-off-by: Hayes Wang <hayeswang@realtek.com>
---
drivers/net/usb/r8152.c | 2 --
1 file changed, 2 deletions(-)
diff --git a/drivers/net/usb/r8152.c b/drivers/net/usb/r8152.c
index ad3abe26b51b..04137ac373b0 100644
--- a/drivers/net/usb/r8152.c
+++ b/drivers/net/usb/r8152.c
@@ -5352,7 +5352,6 @@ static int rtl8152_probe(struct usb_interface *intf,
return 0;
out1:
- netif_napi_del(&tp->napi);
usb_set_intfdata(intf, NULL);
out:
free_netdev(netdev);
@@ -5367,7 +5366,6 @@ static void rtl8152_disconnect(struct usb_interface *intf)
if (tp) {
rtl_set_unplug(tp);
- netif_napi_del(&tp->napi);
unregister_netdev(tp->netdev);
cancel_delayed_work_sync(&tp->hw_phy_work);
tp->rtl_ops.unload(tp);
--
2.21.0
^ permalink raw reply related
* [PATCH net v4 0/2] r8152: fix side effect
From: Hayes Wang @ 2019-08-28 12:56 UTC (permalink / raw)
To: netdev; +Cc: nic_swsd, linux-kernel, Hayes Wang
In-Reply-To: <1394712342-15778-314-Taiwan-albertk@realtek.com>
v4:
Add Fixes tag for both patch #1 and #2.
v3:
Update the commit message for patch #1.
v2:
Replace patch #2 with "r8152: remove calling netif_napi_del".
v1:
The commit 0ee1f4734967 ("r8152: napi hangup fix after disconnect")
add a check to avoid using napi_disable after netif_napi_del. However,
the commit ffa9fec30ca0 ("r8152: set RTL8152_UNPLUG only for real
disconnection") let the check useless.
Therefore, I revert commit 0ee1f4734967 ("r8152: napi hangup fix
after disconnect") first, and add another patch to fix it.
Hayes Wang (2):
Revert "r8152: napi hangup fix after disconnect"
r8152: remove calling netif_napi_del
drivers/net/usb/r8152.c | 5 +----
1 file changed, 1 insertion(+), 4 deletions(-)
--
2.21.0
^ permalink raw reply
* Re: [PATCH 1/2] PTP: introduce new versions of IOCTLs
From: Richard Cochran @ 2019-08-28 12:57 UTC (permalink / raw)
To: Felipe Balbi; +Cc: Joe Perches, Christopher S Hall, netdev, linux-kernel
In-Reply-To: <87k1axwvei.fsf@gmail.com>
On Wed, Aug 28, 2019 at 11:23:33AM +0300, Felipe Balbi wrote:
> Originally I had memset only on the three cases where they were
> needed. Richard, which do you prefer? I don't mind changing it back.
Go ahead and change it back.
Thanks,
Richard
^ permalink raw reply
* Re: [PATCH REPOST 1/2] can: flexcan: fix deadlock when using self wakeup
From: Sean Nyekjaer @ 2019-08-28 13:24 UTC (permalink / raw)
To: Joakim Zhang, mkl@pengutronix.de, linux-can@vger.kernel.org
Cc: wg@grandegger.com, netdev@vger.kernel.org, dl-linux-imx,
Martin Hundebøll
In-Reply-To: <35190c5b-f8be-8784-5b4f-32a691a6cffe@geanix.com>
On 20/08/2019 13.55, Sean Nyekjaer wrote:
>
> I have added some more debug, same test setup:
> https://gist.github.com/sknsean/81208714de23aa3639d3e31dccb2f3e0
>
> root@iwg26:~# systemctl suspend
>
> ...
> https://gist.github.com/sknsean/2a786f1543305056d4de03d387872403
>
> /Sean
Any luck reproducing this?
/Sean
^ permalink raw reply
* RE: [PATCH net-next v2 3/3] dpaa2-eth: Add pause frame support
From: Ioana Ciocoi Radulescu @ 2019-08-28 13:31 UTC (permalink / raw)
To: Andrew Lunn; +Cc: netdev@vger.kernel.org, davem@davemloft.net, Ioana Ciornei
In-Reply-To: <20190828115250.GA32178@lunn.ch>
> -----Original Message-----
> From: Andrew Lunn <andrew@lunn.ch>
> Sent: Wednesday, August 28, 2019 2:53 PM
> To: Ioana Ciocoi Radulescu <ruxandra.radulescu@nxp.com>
> Cc: netdev@vger.kernel.org; davem@davemloft.net; Ioana Ciornei
> <ioana.ciornei@nxp.com>
> Subject: Re: [PATCH net-next v2 3/3] dpaa2-eth: Add pause frame support
>
> > Clearing the ASYM_PAUSE flag only means we tell the firmware we want
> > both Rx and Tx pause to be enabled in the beginning. User can still set
> > an asymmetric config (i.e. only Rx pause or only Tx pause to be enabled)
> > if needed.
> >
> > The truth table is like this:
> >
> > PAUSE | ASYM_PAUSE | Rx pause | Tx pause
> > ----------------------------------------
> > 0 | 0 | disabled | disabled
> > 0 | 1 | disabled | enabled
> > 1 | 0 | enabled | enabled
> > 1 | 1 | enabled | disabled
>
> Hi Ioana
>
> Ah, that is not intuitive. Please add a comment, and maybe this table
> to the commit message.
I think firmware tried to mirror the ASM_DIR bit (see
http://www.ieee802.org/3/z/public/presentations/nov1996/asym.pdf),
but I agree it's not really user friendly. Will add comment in v3.
Thanks,
Ioana
^ permalink raw reply
* [PATCH net-next 11/15] net: sgi: ioc3-eth: use dma-direct for dma allocations
From: Thomas Bogendoerfer @ 2019-08-28 14:03 UTC (permalink / raw)
To: Ralf Baechle, Paul Burton, James Hogan, David S. Miller,
linux-mips, linux-kernel, netdev
In-Reply-To: <20190828140315.17048-1-tbogendoerfer@suse.de>
Replace the homegrown DMA memory allocation, which only works on
SGI-IP27 machines, with the generic dma allocations.
Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 146 ++++++++++++++++++++++++++++--------
1 file changed, 114 insertions(+), 32 deletions(-)
diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index ecd3bc135cc0..aea3fecfac24 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -36,7 +36,6 @@
#include <linux/ip.h>
#include <linux/tcp.h>
#include <linux/udp.h>
-#include <linux/dma-mapping.h>
#include <linux/gfp.h>
#ifdef CONFIG_SERIAL_8250
@@ -49,6 +48,8 @@
#include <linux/etherdevice.h>
#include <linux/ethtool.h>
#include <linux/skbuff.h>
+#include <linux/dma-direct.h>
+
#include <net/ip.h>
#include <asm/byteorder.h>
@@ -64,10 +65,12 @@
#define RX_BUFFS 64
#define RX_RING_ENTRIES 512 /* fixed in hardware */
#define RX_RING_MASK (RX_RING_ENTRIES - 1)
+#define RX_RING_SIZE (RX_RING_ENTRIES * sizeof(u64))
/* 128 TX buffers (not tunable) */
#define TX_RING_ENTRIES 128
#define TX_RING_MASK (TX_RING_ENTRIES - 1)
+#define TX_RING_SIZE (TX_RING_ENTRIES * sizeof(struct ioc3_etxd))
/* IOC3 does dma transfers in 128 byte blocks */
#define IOC3_DMA_XFER_LEN 128UL
@@ -83,9 +86,12 @@
struct ioc3_private {
struct ioc3_ethregs *regs;
struct ioc3 *all_regs;
+ struct device *dma_dev;
u32 *ssram;
unsigned long *rxr; /* pointer to receiver ring */
struct ioc3_etxd *txr;
+ dma_addr_t rxr_dma;
+ dma_addr_t txr_dma;
struct sk_buff *rx_skbs[RX_RING_ENTRIES];
struct sk_buff *tx_skbs[TX_RING_ENTRIES];
int rx_ci; /* RX consumer index */
@@ -125,9 +131,11 @@ static inline unsigned long aligned_rx_skb_addr(unsigned long addr)
return (~addr + 1) & (IOC3_DMA_XFER_LEN - 1UL);
}
-static inline int ioc3_alloc_skb(struct sk_buff **skb, struct ioc3_erxbuf **rxb)
+static inline int ioc3_alloc_skb(struct ioc3_private *ip, struct sk_buff **skb,
+ struct ioc3_erxbuf **rxb, dma_addr_t *rxb_dma)
{
struct sk_buff *new_skb;
+ dma_addr_t d;
int offset;
new_skb = alloc_skb(RX_BUF_SIZE + IOC3_DMA_XFER_LEN - 1, GFP_ATOMIC);
@@ -139,6 +147,14 @@ static inline int ioc3_alloc_skb(struct sk_buff **skb, struct ioc3_erxbuf **rxb)
if (offset)
skb_reserve(new_skb, offset);
+ d = dma_map_single(ip->dma_dev, new_skb->data,
+ RX_BUF_SIZE, DMA_FROM_DEVICE);
+
+ if (dma_mapping_error(ip->dma_dev, d)) {
+ dev_kfree_skb_any(new_skb);
+ return -ENOMEM;
+ }
+ *rxb_dma = d;
*rxb = (struct ioc3_erxbuf *)new_skb->data;
skb_reserve(new_skb, RX_OFFSET);
*skb = new_skb;
@@ -146,17 +162,22 @@ static inline int ioc3_alloc_skb(struct sk_buff **skb, struct ioc3_erxbuf **rxb)
return 0;
}
-static inline unsigned long ioc3_map(void *ptr, unsigned long vdev)
+#ifdef CONFIG_PCI_XTALK_BRIDGE
+static inline unsigned long ioc3_map(dma_addr_t addr, unsigned long attr)
{
-#ifdef CONFIG_SGI_IP27
- vdev <<= 57; /* Shift to PCI64_ATTR_VIRTUAL */
+ return (addr & ~PCI64_ATTR_BAR) | attr;
+}
- return vdev | (0xaUL << PCI64_ATTR_TARG_SHFT) | PCI64_ATTR_PREF |
- ((unsigned long)ptr & TO_PHYS_MASK);
+#define ERBAR_VAL (ERBAR_BARRIER_BIT << ERBAR_RXBARR_SHIFT)
#else
- return virt_to_bus(ptr);
-#endif
+static inline unsigned long ioc3_map(dma_addr_t addr, unsigned long attr)
+{
+ return addr;
}
+
+#define ERBAR_VAL 0
+#endif
+
#define IOC3_SIZE 0x100000
static inline u32 mcr_pack(u32 pulse, u32 sample)
@@ -523,6 +544,7 @@ static inline void ioc3_rx(struct net_device *dev)
int rx_entry, n_entry, len;
struct ioc3_erxbuf *rxb;
unsigned long *rxr;
+ dma_addr_t d;
u32 w0, err;
rxr = ip->rxr; /* Ring base */
@@ -540,12 +562,13 @@ static inline void ioc3_rx(struct net_device *dev)
skb_put(skb, len);
skb->protocol = eth_type_trans(skb, dev);
- if (ioc3_alloc_skb(&new_skb, &rxb)) {
+ if (ioc3_alloc_skb(ip, &new_skb, &rxb, &d)) {
/* Ouch, drop packet and just recycle packet
* to keep the ring filled.
*/
dev->stats.rx_dropped++;
new_skb = skb;
+ d = rxr[rx_entry];
goto next;
}
@@ -554,6 +577,9 @@ static inline void ioc3_rx(struct net_device *dev)
w0 & ERXBUF_IPCKSUM_MASK,
len);
+ dma_unmap_single(ip->dma_dev, rxr[rx_entry],
+ RX_BUF_SIZE, DMA_FROM_DEVICE);
+
netif_rx(skb);
ip->rx_skbs[rx_entry] = NULL; /* Poison */
@@ -566,15 +592,17 @@ static inline void ioc3_rx(struct net_device *dev)
* recycle it.
*/
new_skb = skb;
+ d = rxr[rx_entry];
dev->stats.rx_errors++;
}
if (err & ERXBUF_CRCERR) /* Statistics */
dev->stats.rx_crc_errors++;
if (err & ERXBUF_FRAMERR)
dev->stats.rx_frame_errors++;
+
next:
ip->rx_skbs[n_entry] = new_skb;
- rxr[n_entry] = cpu_to_be64(ioc3_map(rxb, 1));
+ rxr[n_entry] = cpu_to_be64(ioc3_map(d, PCI64_ATTR_BAR));
rxb->w0 = 0; /* Clear valid flag */
n_entry = (n_entry + 1) & RX_RING_MASK; /* Update erpir */
@@ -767,6 +795,26 @@ static void ioc3_mii_start(struct ioc3_private *ip)
add_timer(&ip->ioc3_timer);
}
+static inline void ioc3_tx_unmap(struct ioc3_private *ip, int entry)
+{
+ struct ioc3_etxd *desc;
+ u32 cmd, bufcnt, len;
+
+ desc = &ip->txr[entry];
+ cmd = be32_to_cpu(desc->cmd);
+ bufcnt = be32_to_cpu(desc->bufcnt);
+ if (cmd & ETXD_B1V) {
+ len = (bufcnt & ETXD_B1CNT_MASK) >> ETXD_B1CNT_SHIFT;
+ dma_unmap_single(ip->dma_dev, be64_to_cpu(desc->p1),
+ len, DMA_TO_DEVICE);
+ }
+ if (cmd & ETXD_B2V) {
+ len = (bufcnt & ETXD_B2CNT_MASK) >> ETXD_B2CNT_SHIFT;
+ dma_unmap_single(ip->dma_dev, be64_to_cpu(desc->p2),
+ len, DMA_TO_DEVICE);
+ }
+}
+
static inline void ioc3_clean_tx_ring(struct ioc3_private *ip)
{
struct sk_buff *skb;
@@ -775,6 +823,7 @@ static inline void ioc3_clean_tx_ring(struct ioc3_private *ip)
for (i = 0; i < TX_RING_ENTRIES; i++) {
skb = ip->tx_skbs[i];
if (skb) {
+ ioc3_tx_unmap(ip, i);
ip->tx_skbs[i] = NULL;
dev_kfree_skb_any(skb);
}
@@ -794,8 +843,12 @@ static void ioc3_free_rx_bufs(struct ioc3_private *ip)
while (n_entry != rx_entry) {
skb = ip->rx_skbs[n_entry];
- if (skb)
+ if (skb) {
+ dma_unmap_single(ip->dma_dev,
+ be64_to_cpu(ip->rxr[n_entry]),
+ RX_BUF_SIZE, DMA_FROM_DEVICE);
dev_kfree_skb_any(skb);
+ }
n_entry = (n_entry + 1) & RX_RING_MASK;
}
@@ -805,6 +858,7 @@ static int ioc3_alloc_rx_bufs(struct net_device *dev)
{
struct ioc3_private *ip = netdev_priv(dev);
struct ioc3_erxbuf *rxb;
+ dma_addr_t d;
int i;
/* Now the rx buffers. The RX ring may be larger but
@@ -812,10 +866,10 @@ static int ioc3_alloc_rx_bufs(struct net_device *dev)
* this for performance and memory later.
*/
for (i = 0; i < RX_BUFFS; i++) {
- if (ioc3_alloc_skb(&ip->rx_skbs[i], &rxb))
+ if (ioc3_alloc_skb(ip, &ip->rx_skbs[i], &rxb, &d))
return -ENOMEM;
- ip->rxr[i] = cpu_to_be64(ioc3_map(rxb, 1));
+ ip->rxr[i] = cpu_to_be64(ioc3_map(d, PCI64_ATTR_BAR));
}
ip->rx_ci = 0;
ip->rx_pi = RX_BUFFS;
@@ -861,13 +915,7 @@ static void ioc3_init(struct net_device *dev)
readl(®s->emcr);
/* Misc registers */
-#ifdef CONFIG_SGI_IP27
- /* Barrier on last store */
- writel(PCI64_ATTR_BAR >> 32, ®s->erbar);
-#else
- /* Let PCI API get it right */
- writel(0, ®s->erbar);
-#endif
+ writel(ERBAR_VAL, ®s->erbar);
readl(®s->etcdc); /* Clear on read */
writel(15, ®s->ercsr); /* RX low watermark */
writel(0, ®s->ertr); /* Interrupt immediately */
@@ -883,13 +931,13 @@ static void ioc3_start(struct ioc3_private *ip)
unsigned long ring;
/* Now the rx ring base, consume & produce registers. */
- ring = ioc3_map(ip->rxr, 0);
+ ring = ioc3_map(ip->rxr_dma, PCI64_ATTR_PREC);
writel(ring >> 32, ®s->erbr_h);
writel(ring & 0xffffffff, ®s->erbr_l);
writel(ip->rx_ci << 3, ®s->ercir);
writel((ip->rx_pi << 3) | ERPIR_ARM, ®s->erpir);
- ring = ioc3_map(ip->txr, 0);
+ ring = ioc3_map(ip->txr_dma, PCI64_ATTR_PREC);
ip->txqlen = 0; /* nothing queued */
@@ -1163,6 +1211,7 @@ static int ioc3_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
ip = netdev_priv(dev);
ip->dev = dev;
+ ip->dma_dev = &pdev->dev;
dev->irq = pdev->irq;
@@ -1189,7 +1238,8 @@ static int ioc3_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
ioc3_stop(ip);
/* Allocate and rx ring. 4kb = 512 entries */
- ip->rxr = (unsigned long *)get_zeroed_page(GFP_ATOMIC);
+ ip->rxr = dma_direct_alloc_pages(ip->dma_dev, RX_RING_SIZE,
+ &ip->rxr_dma, GFP_ATOMIC, 0);
if (!ip->rxr) {
pr_err("ioc3-eth: rx ring allocation failed\n");
err = -ENOMEM;
@@ -1197,7 +1247,9 @@ static int ioc3_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
}
/* Allocate tx rings. 16kb = 128 bufs. */
- ip->txr = (struct ioc3_etxd *)__get_free_pages(GFP_KERNEL, 2);
+ ip->txr = dma_direct_alloc_pages(ip->dma_dev, TX_RING_SIZE,
+ &ip->txr_dma,
+ GFP_KERNEL | __GFP_ZERO, 0);
if (!ip->txr) {
pr_err("ioc3-eth: tx ring allocation failed\n");
err = -ENOMEM;
@@ -1256,8 +1308,12 @@ static int ioc3_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
out_stop:
del_timer_sync(&ip->ioc3_timer);
- kfree(ip->rxr);
- kfree(ip->txr);
+ if (ip->rxr)
+ dma_direct_free_pages(ip->dma_dev, RX_RING_SIZE, ip->rxr,
+ ip->rxr_dma, 0);
+ if (ip->txr)
+ dma_direct_free_pages(ip->dma_dev, TX_RING_SIZE, ip->txr,
+ ip->txr_dma, 0);
out_res:
pci_release_regions(pdev);
out_free:
@@ -1275,8 +1331,12 @@ static void ioc3_remove_one(struct pci_dev *pdev)
struct net_device *dev = pci_get_drvdata(pdev);
struct ioc3_private *ip = netdev_priv(dev);
- kfree(ip->rxr);
- kfree(ip->txr);
+ if (ip->rxr)
+ dma_direct_free_pages(ip->dma_dev, RX_RING_SIZE, ip->rxr,
+ ip->rxr_dma, 0);
+ if (ip->txr)
+ dma_direct_free_pages(ip->dma_dev, TX_RING_SIZE, ip->txr,
+ ip->txr_dma, 0);
unregister_netdev(dev);
del_timer_sync(&ip->ioc3_timer);
@@ -1382,18 +1442,32 @@ static netdev_tx_t ioc3_start_xmit(struct sk_buff *skb, struct net_device *dev)
unsigned long b2 = (data | 0x3fffUL) + 1UL;
unsigned long s1 = b2 - data;
unsigned long s2 = data + len - b2;
+ dma_addr_t d1, d2;
desc->cmd = cpu_to_be32(len | ETXD_INTWHENDONE |
ETXD_B1V | ETXD_B2V | w0);
desc->bufcnt = cpu_to_be32((s1 << ETXD_B1CNT_SHIFT) |
(s2 << ETXD_B2CNT_SHIFT));
- desc->p1 = cpu_to_be64(ioc3_map(skb->data, 1));
- desc->p2 = cpu_to_be64(ioc3_map((void *)b2, 1));
+ d1 = dma_map_single(ip->dma_dev, skb->data, s1, DMA_TO_DEVICE);
+ if (dma_mapping_error(ip->dma_dev, d1))
+ goto drop_packet;
+ d2 = dma_map_single(ip->dma_dev, (void *)b2, s1, DMA_TO_DEVICE);
+ if (dma_mapping_error(ip->dma_dev, d2)) {
+ dma_unmap_single(ip->dma_dev, d1, len, DMA_TO_DEVICE);
+ goto drop_packet;
+ }
+ desc->p1 = cpu_to_be64(ioc3_map(d1, PCI64_ATTR_PREF));
+ desc->p2 = cpu_to_be64(ioc3_map(d2, PCI64_ATTR_PREF));
} else {
+ dma_addr_t d;
+
/* Normal sized packet that doesn't cross a page boundary. */
desc->cmd = cpu_to_be32(len | ETXD_INTWHENDONE | ETXD_B1V | w0);
desc->bufcnt = cpu_to_be32(len << ETXD_B1CNT_SHIFT);
- desc->p1 = cpu_to_be64(ioc3_map(skb->data, 1));
+ d = dma_map_single(ip->dma_dev, skb->data, len, DMA_TO_DEVICE);
+ if (dma_mapping_error(ip->dma_dev, d))
+ goto drop_packet;
+ desc->p1 = cpu_to_be64(ioc3_map(d, PCI64_ATTR_PREF));
}
mb(); /* make sure all descriptor changes are visible */
@@ -1411,6 +1485,14 @@ static netdev_tx_t ioc3_start_xmit(struct sk_buff *skb, struct net_device *dev)
spin_unlock_irq(&ip->ioc3_lock);
return NETDEV_TX_OK;
+
+drop_packet:
+ dev_kfree_skb_any(skb);
+ dev->stats.tx_dropped++;
+
+ spin_unlock_irq(&ip->ioc3_lock);
+
+ return NETDEV_TX_OK;
}
static void ioc3_timeout(struct net_device *dev)
--
2.13.7
^ permalink raw reply related
* [PATCH net-next 08/15] net: sgi: ioc3-eth: introduce chip start function
From: Thomas Bogendoerfer @ 2019-08-28 14:03 UTC (permalink / raw)
To: Ralf Baechle, Paul Burton, James Hogan, David S. Miller,
linux-mips, linux-kernel, netdev
In-Reply-To: <20190828140315.17048-1-tbogendoerfer@suse.de>
ioc3_init did everything from reset to init rings to starting the chip.
This change move out chip start into a new function as preparation
for easier handling of receive buffer allocation failures.
Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 49 ++++++++++++++++++++++---------------
1 file changed, 29 insertions(+), 20 deletions(-)
diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index 6c79be3098c3..191eafbeb455 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -105,6 +105,7 @@ static void ioc3_set_multicast_list(struct net_device *dev);
static netdev_tx_t ioc3_start_xmit(struct sk_buff *skb, struct net_device *dev);
static void ioc3_timeout(struct net_device *dev);
static inline unsigned int ioc3_hash(const unsigned char *addr);
+static void ioc3_start(struct ioc3_private *ip);
static inline void ioc3_stop(struct ioc3_private *ip);
static void ioc3_init(struct net_device *dev);
@@ -660,6 +661,7 @@ static void ioc3_error(struct net_device *dev, u32 eisr)
ioc3_stop(ip);
ioc3_init(dev);
+ ioc3_start(ip);
ioc3_mii_init(ip);
netif_wake_queue(dev);
@@ -829,31 +831,11 @@ static void ioc3_alloc_rx_bufs(struct net_device *dev)
static void ioc3_init_rings(struct net_device *dev)
{
struct ioc3_private *ip = netdev_priv(dev);
- struct ioc3_ethregs *regs = ip->regs;
- unsigned long ring;
ioc3_free_rx_bufs(ip);
ioc3_alloc_rx_bufs(dev);
ioc3_clean_tx_ring(ip);
-
- /* Now the rx ring base, consume & produce registers. */
- ring = ioc3_map(ip->rxr, 0);
- writel(ring >> 32, ®s->erbr_h);
- writel(ring & 0xffffffff, ®s->erbr_l);
- writel(ip->rx_ci << 3, ®s->ercir);
- writel((ip->rx_pi << 3) | ERPIR_ARM, ®s->erpir);
-
- ring = ioc3_map(ip->txr, 0);
-
- ip->txqlen = 0; /* nothing queued */
-
- /* Now the tx ring base, consume & produce registers. */
- writel(ring >> 32, ®s->etbr_h);
- writel(ring & 0xffffffff, ®s->etbr_l);
- writel(ip->tx_pi << 7, ®s->etpir);
- writel(ip->tx_ci << 7, ®s->etcir);
- readl(®s->etcir); /* Flush */
}
static inline void ioc3_ssram_disc(struct ioc3_private *ip)
@@ -910,6 +892,30 @@ static void ioc3_init(struct net_device *dev)
writel(42, ®s->ersr); /* XXX should be random */
ioc3_init_rings(dev);
+}
+
+static void ioc3_start(struct ioc3_private *ip)
+{
+ struct ioc3_ethregs *regs = ip->regs;
+ unsigned long ring;
+
+ /* Now the rx ring base, consume & produce registers. */
+ ring = ioc3_map(ip->rxr, 0);
+ writel(ring >> 32, ®s->erbr_h);
+ writel(ring & 0xffffffff, ®s->erbr_l);
+ writel(ip->rx_ci << 3, ®s->ercir);
+ writel((ip->rx_pi << 3) | ERPIR_ARM, ®s->erpir);
+
+ ring = ioc3_map(ip->txr, 0);
+
+ ip->txqlen = 0; /* nothing queued */
+
+ /* Now the tx ring base, consume & produce registers. */
+ writel(ring >> 32, ®s->etbr_h);
+ writel(ring & 0xffffffff, ®s->etbr_l);
+ writel(ip->tx_pi << 7, ®s->etpir);
+ writel(ip->tx_ci << 7, ®s->etcir);
+ readl(®s->etcir); /* Flush */
ip->emcr |= ((RX_OFFSET / 2) << EMCR_RXOFF_SHIFT) | EMCR_TXDMAEN |
EMCR_TXEN | EMCR_RXDMAEN | EMCR_RXEN | EMCR_PADEN;
@@ -942,6 +948,7 @@ static int ioc3_open(struct net_device *dev)
ip->ehar_h = 0;
ip->ehar_l = 0;
ioc3_init(dev);
+ ioc3_start(ip);
ioc3_mii_start(ip);
netif_start_queue(dev);
@@ -1210,6 +1217,7 @@ static int ioc3_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
}
ioc3_init(dev);
+ ioc3_start(ip);
ip->pdev = pdev;
@@ -1430,6 +1438,7 @@ static void ioc3_timeout(struct net_device *dev)
ioc3_stop(ip);
ioc3_init(dev);
+ ioc3_start(ip);
ioc3_mii_init(ip);
ioc3_mii_start(ip);
--
2.13.7
^ permalink raw reply related
* [PATCH net-next 15/15] net: sgi: ioc3-eth: no need to stop queue set_multicast_list
From: Thomas Bogendoerfer @ 2019-08-28 14:03 UTC (permalink / raw)
To: Ralf Baechle, Paul Burton, James Hogan, David S. Miller,
linux-mips, linux-kernel, netdev
In-Reply-To: <20190828140315.17048-1-tbogendoerfer@suse.de>
netif_stop_queue()/netif_wake_qeue() aren't needed for changing
multicast filters.
Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 4 ----
1 file changed, 4 deletions(-)
diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index 5de8300e5eb2..aaa31ee9c8da 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -1629,8 +1629,6 @@ static void ioc3_set_multicast_list(struct net_device *dev)
struct netdev_hw_addr *ha;
u64 ehar = 0;
- netif_stop_queue(dev); /* Lock out others. */
-
spin_lock_irq(&ip->ioc3_lock);
if (dev->flags & IFF_PROMISC) { /* Set promiscuous. */
@@ -1662,8 +1660,6 @@ static void ioc3_set_multicast_list(struct net_device *dev)
}
spin_unlock_irq(&ip->ioc3_lock);
-
- netif_wake_queue(dev); /* Let us get going again. */
}
module_pci_driver(ioc3_driver);
--
2.13.7
^ permalink raw reply related
* [PATCH net-next 14/15] net: sgi: ioc3-eth: protect emcr in all cases
From: Thomas Bogendoerfer @ 2019-08-28 14:03 UTC (permalink / raw)
To: Ralf Baechle, Paul Burton, James Hogan, David S. Miller,
linux-mips, linux-kernel, netdev
In-Reply-To: <20190828140315.17048-1-tbogendoerfer@suse.de>
emcr in private struct wasn't always protected by spinlock.
Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index b8aea19fbeba..5de8300e5eb2 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -729,6 +729,8 @@ static inline void ioc3_setup_duplex(struct ioc3_private *ip)
{
struct ioc3_ethregs *regs = ip->regs;
+ spin_lock_irq(&ip->ioc3_lock);
+
if (ip->mii.full_duplex) {
writel(ETCSR_FD, ®s->etcsr);
ip->emcr |= EMCR_DUPLEX;
@@ -737,6 +739,8 @@ static inline void ioc3_setup_duplex(struct ioc3_private *ip)
ip->emcr &= ~EMCR_DUPLEX;
}
writel(ip->emcr, ®s->emcr);
+
+ spin_unlock_irq(&ip->ioc3_lock);
}
static void ioc3_timer(struct timer_list *t)
@@ -1627,6 +1631,8 @@ static void ioc3_set_multicast_list(struct net_device *dev)
netif_stop_queue(dev); /* Lock out others. */
+ spin_lock_irq(&ip->ioc3_lock);
+
if (dev->flags & IFF_PROMISC) { /* Set promiscuous. */
ip->emcr |= EMCR_PROMISC;
writel(ip->emcr, ®s->emcr);
@@ -1655,6 +1661,8 @@ static void ioc3_set_multicast_list(struct net_device *dev)
writel(ip->ehar_l, ®s->ehar_l);
}
+ spin_unlock_irq(&ip->ioc3_lock);
+
netif_wake_queue(dev); /* Let us get going again. */
}
--
2.13.7
^ permalink raw reply related
* [PATCH net-next 10/15] net: sgi: ioc3-eth: refactor rx buffer allocation
From: Thomas Bogendoerfer @ 2019-08-28 14:03 UTC (permalink / raw)
To: Ralf Baechle, Paul Burton, James Hogan, David S. Miller,
linux-mips, linux-kernel, netdev
In-Reply-To: <20190828140315.17048-1-tbogendoerfer@suse.de>
Move common code for rx buffer setup into ioc3_alloc_skb and deal
with allocation failures. Also clean up allocation size calculation.
Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 95 ++++++++++++++++++-------------------
1 file changed, 45 insertions(+), 50 deletions(-)
diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index e51922b5ca04..ecd3bc135cc0 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -11,11 +11,8 @@
*
* To do:
*
- * o Handle allocation failures in ioc3_alloc_skb() more gracefully.
- * o Handle allocation failures in ioc3_init_rings().
* o Use prefetching for large packets. What is a good lower limit for
* prefetching?
- * o We're probably allocating a bit too much memory.
* o Use hardware checksums.
* o Convert to using a IOC3 meta driver.
* o Which PHYs might possibly be attached to the IOC3 in real live,
@@ -72,6 +69,13 @@
#define TX_RING_ENTRIES 128
#define TX_RING_MASK (TX_RING_ENTRIES - 1)
+/* IOC3 does dma transfers in 128 byte blocks */
+#define IOC3_DMA_XFER_LEN 128UL
+
+/* Every RX buffer starts with 8 byte descriptor data */
+#define RX_OFFSET (sizeof(struct ioc3_erxbuf) + NET_IP_ALIGN)
+#define RX_BUF_SIZE (13 * IOC3_DMA_XFER_LEN)
+
#define ETCSR_FD ((17 << ETCSR_IPGR2_SHIFT) | (11 << ETCSR_IPGR1_SHIFT) | 21)
#define ETCSR_HD ((21 << ETCSR_IPGR2_SHIFT) | (21 << ETCSR_IPGR1_SHIFT) | 21)
@@ -108,36 +112,38 @@ static inline unsigned int ioc3_hash(const unsigned char *addr);
static void ioc3_start(struct ioc3_private *ip);
static inline void ioc3_stop(struct ioc3_private *ip);
static void ioc3_init(struct net_device *dev);
-static void ioc3_alloc_rx_bufs(struct net_device *dev);
+static int ioc3_alloc_rx_bufs(struct net_device *dev);
static void ioc3_free_rx_bufs(struct ioc3_private *ip);
static inline void ioc3_clean_tx_ring(struct ioc3_private *ip);
static const char ioc3_str[] = "IOC3 Ethernet";
static const struct ethtool_ops ioc3_ethtool_ops;
-/* We use this to acquire receive skb's that we can DMA directly into. */
-
-#define IOC3_CACHELINE 128UL
static inline unsigned long aligned_rx_skb_addr(unsigned long addr)
{
- return (~addr + 1) & (IOC3_CACHELINE - 1UL);
+ return (~addr + 1) & (IOC3_DMA_XFER_LEN - 1UL);
}
-static inline struct sk_buff *ioc3_alloc_skb(unsigned long length,
- unsigned int gfp_mask)
+static inline int ioc3_alloc_skb(struct sk_buff **skb, struct ioc3_erxbuf **rxb)
{
- struct sk_buff *skb;
+ struct sk_buff *new_skb;
+ int offset;
- skb = alloc_skb(length + IOC3_CACHELINE - 1, gfp_mask);
- if (likely(skb)) {
- int offset = aligned_rx_skb_addr((unsigned long)skb->data);
+ new_skb = alloc_skb(RX_BUF_SIZE + IOC3_DMA_XFER_LEN - 1, GFP_ATOMIC);
+ if (!new_skb)
+ return -ENOMEM;
- if (offset)
- skb_reserve(skb, offset);
- }
+ /* ensure buffer is aligned to IOC3_DMA_XFER_LEN */
+ offset = aligned_rx_skb_addr((unsigned long)new_skb->data);
+ if (offset)
+ skb_reserve(new_skb, offset);
+
+ *rxb = (struct ioc3_erxbuf *)new_skb->data;
+ skb_reserve(new_skb, RX_OFFSET);
+ *skb = new_skb;
- return skb;
+ return 0;
}
static inline unsigned long ioc3_map(void *ptr, unsigned long vdev)
@@ -151,13 +157,6 @@ static inline unsigned long ioc3_map(void *ptr, unsigned long vdev)
return virt_to_bus(ptr);
#endif
}
-
-/* BEWARE: The IOC3 documentation documents the size of rx buffers as
- * 1644 while it's actually 1664. This one was nasty to track down ...
- */
-#define RX_OFFSET 10
-#define RX_BUF_ALLOC_SIZE (1664 + RX_OFFSET + IOC3_CACHELINE)
-
#define IOC3_SIZE 0x100000
static inline u32 mcr_pack(u32 pulse, u32 sample)
@@ -538,11 +537,10 @@ static inline void ioc3_rx(struct net_device *dev)
err = be32_to_cpu(rxb->err); /* It's valid ... */
if (err & ERXBUF_GOODPKT) {
len = ((w0 >> ERXBUF_BYTECNT_SHIFT) & 0x7ff) - 4;
- skb_trim(skb, len);
+ skb_put(skb, len);
skb->protocol = eth_type_trans(skb, dev);
- new_skb = ioc3_alloc_skb(RX_BUF_ALLOC_SIZE, GFP_ATOMIC);
- if (!new_skb) {
+ if (ioc3_alloc_skb(&new_skb, &rxb)) {
/* Ouch, drop packet and just recycle packet
* to keep the ring filled.
*/
@@ -560,11 +558,6 @@ static inline void ioc3_rx(struct net_device *dev)
ip->rx_skbs[rx_entry] = NULL; /* Poison */
- /* Because we reserve afterwards. */
- skb_put(new_skb, (1664 + RX_OFFSET));
- rxb = (struct ioc3_erxbuf *)new_skb->data;
- skb_reserve(new_skb, RX_OFFSET);
-
dev->stats.rx_packets++; /* Statistics */
dev->stats.rx_bytes += len;
} else {
@@ -667,7 +660,11 @@ static void ioc3_error(struct net_device *dev, u32 eisr)
ioc3_clean_tx_ring(ip);
ioc3_init(dev);
- ioc3_alloc_rx_bufs(dev);
+ if (ioc3_alloc_rx_bufs(dev)) {
+ netdev_err(dev, "ioc3_error: rx buffer allocation failed\n");
+ spin_unlock(&ip->ioc3_lock);
+ return;
+ }
ioc3_start(ip);
ioc3_mii_init(ip);
@@ -804,7 +801,7 @@ static void ioc3_free_rx_bufs(struct ioc3_private *ip)
}
}
-static void ioc3_alloc_rx_bufs(struct net_device *dev)
+static int ioc3_alloc_rx_bufs(struct net_device *dev)
{
struct ioc3_private *ip = netdev_priv(dev);
struct ioc3_erxbuf *rxb;
@@ -815,24 +812,15 @@ static void ioc3_alloc_rx_bufs(struct net_device *dev)
* this for performance and memory later.
*/
for (i = 0; i < RX_BUFFS; i++) {
- struct sk_buff *skb;
-
- skb = ioc3_alloc_skb(RX_BUF_ALLOC_SIZE, GFP_ATOMIC);
- if (!skb) {
- show_free_areas(0, NULL);
- continue;
- }
-
- ip->rx_skbs[i] = skb;
+ if (ioc3_alloc_skb(&ip->rx_skbs[i], &rxb))
+ return -ENOMEM;
- /* Because we reserve afterwards. */
- skb_put(skb, (1664 + RX_OFFSET));
- rxb = (struct ioc3_erxbuf *)skb->data;
ip->rxr[i] = cpu_to_be64(ioc3_map(rxb, 1));
- skb_reserve(skb, RX_OFFSET);
}
ip->rx_ci = 0;
ip->rx_pi = RX_BUFFS;
+
+ return 0;
}
static inline void ioc3_ssram_disc(struct ioc3_private *ip)
@@ -944,7 +932,10 @@ static int ioc3_open(struct net_device *dev)
ip->ehar_l = 0;
ioc3_init(dev);
- ioc3_alloc_rx_bufs(dev);
+ if (ioc3_alloc_rx_bufs(dev)) {
+ netdev_err(dev, "ioc3_open: rx buffer allocation failed\n");
+ return -ENOMEM;
+ }
ioc3_start(ip);
ioc3_mii_start(ip);
@@ -1435,7 +1426,11 @@ static void ioc3_timeout(struct net_device *dev)
ioc3_clean_tx_ring(ip);
ioc3_init(dev);
- ioc3_alloc_rx_bufs(dev);
+ if (ioc3_alloc_rx_bufs(dev)) {
+ netdev_err(dev, "ioc3_timeout: rx buffer allocation failed\n");
+ spin_unlock_irq(&ip->ioc3_lock);
+ return;
+ }
ioc3_start(ip);
ioc3_mii_init(ip);
ioc3_mii_start(ip);
--
2.13.7
^ permalink raw reply related
* [PATCH net-next 12/15] net: sgi: ioc3-eth: use csum_fold
From: Thomas Bogendoerfer @ 2019-08-28 14:03 UTC (permalink / raw)
To: Ralf Baechle, Paul Burton, James Hogan, David S. Miller,
linux-mips, linux-kernel, netdev
In-Reply-To: <20190828140315.17048-1-tbogendoerfer@suse.de>
replace open coded checksum folding by csum_fold.
Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 6 +-----
1 file changed, 1 insertion(+), 5 deletions(-)
diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index aea3fecfac24..82a298eea8d4 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -1393,16 +1393,12 @@ static netdev_tx_t ioc3_start_xmit(struct sk_buff *skb, struct net_device *dev)
/* Sum up dest addr, src addr and protocol */
ehsum = eh[0] + eh[1] + eh[2] + eh[3] + eh[4] + eh[5] + eh[6];
- /* Fold ehsum. can't use csum_fold which negates also ... */
- ehsum = (ehsum & 0xffff) + (ehsum >> 16);
- ehsum = (ehsum & 0xffff) + (ehsum >> 16);
-
/* Skip IP header; it's sum is always zero and was
* already filled in by ip_output.c
*/
csum = csum_tcpudp_nofold(ih->saddr, ih->daddr,
ih->tot_len - (ih->ihl << 2),
- proto, 0xffff ^ ehsum);
+ proto, csum_fold(ehsum));
csum = (csum & 0xffff) + (csum >> 16); /* Fold again */
csum = (csum & 0xffff) + (csum >> 16);
--
2.13.7
^ permalink raw reply related
* [PATCH net-next 09/15] net: sgi: ioc3-eth: split ring cleaning/freeing and allocation
From: Thomas Bogendoerfer @ 2019-08-28 14:03 UTC (permalink / raw)
To: Ralf Baechle, Paul Burton, James Hogan, David S. Miller,
linux-mips, linux-kernel, netdev
In-Reply-To: <20190828140315.17048-1-tbogendoerfer@suse.de>
Do tx ring cleaning and freeing of rx buffers, when chip is shutdown and
allocate buffers before bringing chip up.
Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 28 +++++++++++++---------------
1 file changed, 13 insertions(+), 15 deletions(-)
diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index 191eafbeb455..e51922b5ca04 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -108,6 +108,9 @@ static inline unsigned int ioc3_hash(const unsigned char *addr);
static void ioc3_start(struct ioc3_private *ip);
static inline void ioc3_stop(struct ioc3_private *ip);
static void ioc3_init(struct net_device *dev);
+static void ioc3_alloc_rx_bufs(struct net_device *dev);
+static void ioc3_free_rx_bufs(struct ioc3_private *ip);
+static inline void ioc3_clean_tx_ring(struct ioc3_private *ip);
static const char ioc3_str[] = "IOC3 Ethernet";
static const struct ethtool_ops ioc3_ethtool_ops;
@@ -660,7 +663,11 @@ static void ioc3_error(struct net_device *dev, u32 eisr)
netdev_err(dev, "TX PCI error.\n");
ioc3_stop(ip);
+ ioc3_free_rx_bufs(ip);
+ ioc3_clean_tx_ring(ip);
+
ioc3_init(dev);
+ ioc3_alloc_rx_bufs(dev);
ioc3_start(ip);
ioc3_mii_init(ip);
@@ -828,16 +835,6 @@ static void ioc3_alloc_rx_bufs(struct net_device *dev)
ip->rx_pi = RX_BUFFS;
}
-static void ioc3_init_rings(struct net_device *dev)
-{
- struct ioc3_private *ip = netdev_priv(dev);
-
- ioc3_free_rx_bufs(ip);
- ioc3_alloc_rx_bufs(dev);
-
- ioc3_clean_tx_ring(ip);
-}
-
static inline void ioc3_ssram_disc(struct ioc3_private *ip)
{
struct ioc3_ethregs *regs = ip->regs;
@@ -890,8 +887,6 @@ static void ioc3_init(struct net_device *dev)
writel(ip->ehar_h, ®s->ehar_h);
writel(ip->ehar_l, ®s->ehar_l);
writel(42, ®s->ersr); /* XXX should be random */
-
- ioc3_init_rings(dev);
}
static void ioc3_start(struct ioc3_private *ip)
@@ -947,7 +942,9 @@ static int ioc3_open(struct net_device *dev)
ip->ehar_h = 0;
ip->ehar_l = 0;
+
ioc3_init(dev);
+ ioc3_alloc_rx_bufs(dev);
ioc3_start(ip);
ioc3_mii_start(ip);
@@ -1217,7 +1214,6 @@ static int ioc3_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
}
ioc3_init(dev);
- ioc3_start(ip);
ip->pdev = pdev;
@@ -1268,9 +1264,7 @@ static int ioc3_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
return 0;
out_stop:
- ioc3_stop(ip);
del_timer_sync(&ip->ioc3_timer);
- ioc3_free_rx_bufs(ip);
kfree(ip->rxr);
kfree(ip->txr);
out_res:
@@ -1437,7 +1431,11 @@ static void ioc3_timeout(struct net_device *dev)
spin_lock_irq(&ip->ioc3_lock);
ioc3_stop(ip);
+ ioc3_free_rx_bufs(ip);
+ ioc3_clean_tx_ring(ip);
+
ioc3_init(dev);
+ ioc3_alloc_rx_bufs(dev);
ioc3_start(ip);
ioc3_mii_init(ip);
ioc3_mii_start(ip);
--
2.13.7
^ permalink raw reply related
* [PATCH net-next 13/15] net: sgi: ioc3-eth: Fix IPG settings
From: Thomas Bogendoerfer @ 2019-08-28 14:03 UTC (permalink / raw)
To: Ralf Baechle, Paul Burton, James Hogan, David S. Miller,
linux-mips, linux-kernel, netdev
In-Reply-To: <20190828140315.17048-1-tbogendoerfer@suse.de>
The half/full duplex settings for inter packet gap counters/timer were
reversed.
Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index 82a298eea8d4..b8aea19fbeba 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -79,8 +79,8 @@
#define RX_OFFSET (sizeof(struct ioc3_erxbuf) + NET_IP_ALIGN)
#define RX_BUF_SIZE (13 * IOC3_DMA_XFER_LEN)
-#define ETCSR_FD ((17 << ETCSR_IPGR2_SHIFT) | (11 << ETCSR_IPGR1_SHIFT) | 21)
-#define ETCSR_HD ((21 << ETCSR_IPGR2_SHIFT) | (21 << ETCSR_IPGR1_SHIFT) | 21)
+#define ETCSR_FD ((21 << ETCSR_IPGR2_SHIFT) | (21 << ETCSR_IPGR1_SHIFT) | 21)
+#define ETCSR_HD ((17 << ETCSR_IPGR2_SHIFT) | (11 << ETCSR_IPGR1_SHIFT) | 21)
/* Private per NIC data of the driver. */
struct ioc3_private {
--
2.13.7
^ permalink raw reply related
* [PATCH net-next 01/15] MIPS: SGI-IP27: remove ioc3 ethernet init
From: Thomas Bogendoerfer @ 2019-08-28 14:03 UTC (permalink / raw)
To: Ralf Baechle, Paul Burton, James Hogan, David S. Miller,
linux-mips, linux-kernel, netdev
In-Reply-To: <20190828140315.17048-1-tbogendoerfer@suse.de>
Removed not needed disabling of ethernet interrupts in IP27 platform code.
Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
arch/mips/sgi-ip27/ip27-init.c | 13 -------------
1 file changed, 13 deletions(-)
diff --git a/arch/mips/sgi-ip27/ip27-init.c b/arch/mips/sgi-ip27/ip27-init.c
index 066b33f50bcc..59d5375c9021 100644
--- a/arch/mips/sgi-ip27/ip27-init.c
+++ b/arch/mips/sgi-ip27/ip27-init.c
@@ -130,17 +130,6 @@ cnodeid_t get_compact_nodeid(void)
return NASID_TO_COMPACT_NODEID(get_nasid());
}
-static inline void ioc3_eth_init(void)
-{
- struct ioc3 *ioc3;
- nasid_t nid;
-
- nid = get_nasid();
- ioc3 = (struct ioc3 *) KL_CONFIG_CH_CONS_INFO(nid)->memory_base;
-
- ioc3->eier = 0;
-}
-
extern void ip27_reboot_setup(void);
void __init plat_mem_setup(void)
@@ -182,8 +171,6 @@ void __init plat_mem_setup(void)
panic("Kernel compiled for N mode.");
#endif
- ioc3_eth_init();
-
ioport_resource.start = 0;
ioport_resource.end = ~0UL;
set_io_port_base(IO_BASE);
--
2.13.7
^ permalink raw reply related
* [PATCH net-next 05/15] net: sgi: ioc3-eth: allocate space for desc rings only once
From: Thomas Bogendoerfer @ 2019-08-28 14:03 UTC (permalink / raw)
To: Ralf Baechle, Paul Burton, James Hogan, David S. Miller,
linux-mips, linux-kernel, netdev
In-Reply-To: <20190828140315.17048-1-tbogendoerfer@suse.de>
Memory for descriptor rings are allocated/freed, when interface is
brought up/down. Since the size of the rings is not changeable by
hardware, we now allocate rings now during probe and free it, when
device is removed.
Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 103 ++++++++++++++++++------------------
1 file changed, 51 insertions(+), 52 deletions(-)
diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index c875640926d6..6ca560d4ab79 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -803,25 +803,17 @@ static void ioc3_free_rings(struct ioc3_private *ip)
struct sk_buff *skb;
int rx_entry, n_entry;
- if (ip->txr) {
- ioc3_clean_tx_ring(ip);
- free_pages((unsigned long)ip->txr, 2);
- ip->txr = NULL;
- }
+ ioc3_clean_tx_ring(ip);
- if (ip->rxr) {
- n_entry = ip->rx_ci;
- rx_entry = ip->rx_pi;
+ n_entry = ip->rx_ci;
+ rx_entry = ip->rx_pi;
- while (n_entry != rx_entry) {
- skb = ip->rx_skbs[n_entry];
- if (skb)
- dev_kfree_skb_any(skb);
+ while (n_entry != rx_entry) {
+ skb = ip->rx_skbs[n_entry];
+ if (skb)
+ dev_kfree_skb_any(skb);
- n_entry = (n_entry + 1) & RX_RING_MASK;
- }
- free_page((unsigned long)ip->rxr);
- ip->rxr = NULL;
+ n_entry = (n_entry + 1) & RX_RING_MASK;
}
}
@@ -829,49 +821,34 @@ static void ioc3_alloc_rings(struct net_device *dev)
{
struct ioc3_private *ip = netdev_priv(dev);
struct ioc3_erxbuf *rxb;
- unsigned long *rxr;
int i;
- if (!ip->rxr) {
- /* Allocate and initialize rx ring. 4kb = 512 entries */
- ip->rxr = (unsigned long *)get_zeroed_page(GFP_ATOMIC);
- rxr = ip->rxr;
- if (!rxr)
- pr_err("%s: get_zeroed_page() failed!\n", __func__);
-
- /* Now the rx buffers. The RX ring may be larger but
- * we only allocate 16 buffers for now. Need to tune
- * this for performance and memory later.
- */
- for (i = 0; i < RX_BUFFS; i++) {
- struct sk_buff *skb;
+ /* Now the rx buffers. The RX ring may be larger but
+ * we only allocate 16 buffers for now. Need to tune
+ * this for performance and memory later.
+ */
+ for (i = 0; i < RX_BUFFS; i++) {
+ struct sk_buff *skb;
- skb = ioc3_alloc_skb(RX_BUF_ALLOC_SIZE, GFP_ATOMIC);
- if (!skb) {
- show_free_areas(0, NULL);
- continue;
- }
+ skb = ioc3_alloc_skb(RX_BUF_ALLOC_SIZE, GFP_ATOMIC);
+ if (!skb) {
+ show_free_areas(0, NULL);
+ continue;
+ }
- ip->rx_skbs[i] = skb;
+ ip->rx_skbs[i] = skb;
- /* Because we reserve afterwards. */
- skb_put(skb, (1664 + RX_OFFSET));
- rxb = (struct ioc3_erxbuf *)skb->data;
- rxr[i] = cpu_to_be64(ioc3_map(rxb, 1));
- skb_reserve(skb, RX_OFFSET);
- }
- ip->rx_ci = 0;
- ip->rx_pi = RX_BUFFS;
+ /* Because we reserve afterwards. */
+ skb_put(skb, (1664 + RX_OFFSET));
+ rxb = (struct ioc3_erxbuf *)skb->data;
+ ip->rxr[i] = cpu_to_be64(ioc3_map(rxb, 1));
+ skb_reserve(skb, RX_OFFSET);
}
+ ip->rx_ci = 0;
+ ip->rx_pi = RX_BUFFS;
- if (!ip->txr) {
- /* Allocate and initialize tx rings. 16kb = 128 bufs. */
- ip->txr = (struct ioc3_etxd *)__get_free_pages(GFP_KERNEL, 2);
- if (!ip->txr)
- pr_err("%s: __get_free_pages() failed!\n", __func__);
- ip->tx_pi = 0;
- ip->tx_ci = 0;
- }
+ ip->tx_pi = 0;
+ ip->tx_ci = 0;
}
static void ioc3_init_rings(struct net_device *dev)
@@ -1239,6 +1216,23 @@ static int ioc3_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
timer_setup(&ip->ioc3_timer, ioc3_timer, 0);
ioc3_stop(ip);
+
+ /* Allocate and rx ring. 4kb = 512 entries */
+ ip->rxr = (unsigned long *)get_zeroed_page(GFP_ATOMIC);
+ if (!ip->rxr) {
+ pr_err("ioc3-eth: rx ring allocation failed\n");
+ err = -ENOMEM;
+ goto out_stop;
+ }
+
+ /* Allocate tx rings. 16kb = 128 bufs. */
+ ip->txr = (struct ioc3_etxd *)__get_free_pages(GFP_KERNEL, 2);
+ if (!ip->txr) {
+ pr_err("ioc3-eth: tx ring allocation failed\n");
+ err = -ENOMEM;
+ goto out_stop;
+ }
+
ioc3_init(dev);
ip->pdev = pdev;
@@ -1293,6 +1287,8 @@ static int ioc3_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
ioc3_stop(ip);
del_timer_sync(&ip->ioc3_timer);
ioc3_free_rings(ip);
+ kfree(ip->rxr);
+ kfree(ip->txr);
out_res:
pci_release_regions(pdev);
out_free:
@@ -1310,6 +1306,9 @@ static void ioc3_remove_one(struct pci_dev *pdev)
struct net_device *dev = pci_get_drvdata(pdev);
struct ioc3_private *ip = netdev_priv(dev);
+ kfree(ip->rxr);
+ kfree(ip->txr);
+
unregister_netdev(dev);
del_timer_sync(&ip->ioc3_timer);
--
2.13.7
^ permalink raw reply related
* [PATCH net-next 07/15] net: sgi: ioc3-eth: separate tx and rx ring handling
From: Thomas Bogendoerfer @ 2019-08-28 14:03 UTC (permalink / raw)
To: Ralf Baechle, Paul Burton, James Hogan, David S. Miller,
linux-mips, linux-kernel, netdev
In-Reply-To: <20190828140315.17048-1-tbogendoerfer@suse.de>
After allocation of descriptor memory is now done once in probe
handling of tx ring is completely done by ioc3_clean_tx_ring. So
we remove the remaining tx ring actions out of ioc3_alloc_rings
and ioc3_free_rings and rename it to ioc3_[alloc|free]_rx_bufs
to better describe what they are doing.
Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 19 ++++++++-----------
1 file changed, 8 insertions(+), 11 deletions(-)
diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index 39631e067b71..6c79be3098c3 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -778,13 +778,11 @@ static inline void ioc3_clean_tx_ring(struct ioc3_private *ip)
ip->tx_ci = 0;
}
-static void ioc3_free_rings(struct ioc3_private *ip)
+static void ioc3_free_rx_bufs(struct ioc3_private *ip)
{
struct sk_buff *skb;
int rx_entry, n_entry;
- ioc3_clean_tx_ring(ip);
-
n_entry = ip->rx_ci;
rx_entry = ip->rx_pi;
@@ -797,7 +795,7 @@ static void ioc3_free_rings(struct ioc3_private *ip)
}
}
-static void ioc3_alloc_rings(struct net_device *dev)
+static void ioc3_alloc_rx_bufs(struct net_device *dev)
{
struct ioc3_private *ip = netdev_priv(dev);
struct ioc3_erxbuf *rxb;
@@ -826,9 +824,6 @@ static void ioc3_alloc_rings(struct net_device *dev)
}
ip->rx_ci = 0;
ip->rx_pi = RX_BUFFS;
-
- ip->tx_pi = 0;
- ip->tx_ci = 0;
}
static void ioc3_init_rings(struct net_device *dev)
@@ -837,8 +832,8 @@ static void ioc3_init_rings(struct net_device *dev)
struct ioc3_ethregs *regs = ip->regs;
unsigned long ring;
- ioc3_free_rings(ip);
- ioc3_alloc_rings(dev);
+ ioc3_free_rx_bufs(ip);
+ ioc3_alloc_rx_bufs(dev);
ioc3_clean_tx_ring(ip);
@@ -964,7 +959,9 @@ static int ioc3_close(struct net_device *dev)
ioc3_stop(ip);
free_irq(dev->irq, dev);
- ioc3_free_rings(ip);
+ ioc3_free_rx_bufs(ip);
+ ioc3_clean_tx_ring(ip);
+
return 0;
}
@@ -1265,7 +1262,7 @@ static int ioc3_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
out_stop:
ioc3_stop(ip);
del_timer_sync(&ip->ioc3_timer);
- ioc3_free_rings(ip);
+ ioc3_free_rx_bufs(ip);
kfree(ip->rxr);
kfree(ip->txr);
out_res:
--
2.13.7
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox