* Re: [PATCH net 1/2] Revert "sfc: Use write-combining to reduce TX latency" and follow-ups
From: David Miller @ 2011-09-15 21:25 UTC (permalink / raw)
To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1314914969.2733.15.camel@bwh-desktop>
From: Ben Hutchings <bhutchings@solarflare.com>
Date: Thu, 01 Sep 2011 23:09:29 +0100
> This reverts commits 65f0b417dee94f779ce9b77102b7d73c93723b39,
> d88d6b05fee3cc78e5b0273eb58c31201dcc6b76,
> fcfa060468a4edcf776f0c1211d826d5de1668c1,
> 747df2258b1b9a2e25929ef496262c339c380009 and
> 867955f5682f7157fdafe8670804b9f8ea077bc7.
>
> Depending on the processor model, write-combining may result in
> reordering that the NIC will not tolerate. This typically results
> in a DMA error event and reset by the driver, logged as:
>
> sfc 0000:0e:00.0: eth2: TX DMA Q reports TX_EV_PKT_ERR.
> sfc 0000:0e:00.0: eth2: resetting (ALL)
>
> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
> ---
> Please queue this for 3.0.y as well.
Applied and queued.
^ permalink raw reply
* Re: [PATCH net 2/2] sfc: Use 64-bit writes for TX push where possible
From: David Miller @ 2011-09-15 21:25 UTC (permalink / raw)
To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1314915000.2733.16.camel@bwh-desktop>
From: Ben Hutchings <bhutchings@solarflare.com>
Date: Thu, 01 Sep 2011 23:09:59 +0100
> This was originally done as part of commit
> 65f0b417dee94f779ce9b77102b7d73c93723b39 ("sfc: Use write-combining to
> reduce TX latency"), but that had to be reverted.
>
> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Applied.
^ permalink raw reply
* Re: [PATCH 1/3] pch_gbe: fixed the issue which receives an unnecessary packet.
From: David Miller @ 2011-09-15 21:25 UTC (permalink / raw)
To: toshiharu-linux
Cc: netdev, linux-kernel, eric.dumazet, qi.wang, yong.y.wang,
joel.clark, kok.howg.ewe, tomoya-linux
In-Reply-To: <1314922809-4041-1-git-send-email-toshiharu-linux@dsn.okisemi.com>
From: Toshiharu Okada <toshiharu-linux@dsn.okisemi.com>
Date: Fri, 2 Sep 2011 09:20:07 +0900
> This patch fixed the issue which receives an unnecessary packet before link
>
> When using PHY of GMII, an unnecessary packet is received,
> And it becomes impossible to receive a packet after link up.
>
> Signed-off-by: Toshiharu Okada <toshiharu-linux@dsn.okisemi.com>
Applied.
^ permalink raw reply
* Re: [PATCH 2/3] pch_gbe: added the process of FIFO over run error
From: David Miller @ 2011-09-15 21:26 UTC (permalink / raw)
To: toshiharu-linux
Cc: netdev, linux-kernel, eric.dumazet, qi.wang, yong.y.wang,
joel.clark, kok.howg.ewe, tomoya-linux
In-Reply-To: <1314922809-4041-2-git-send-email-toshiharu-linux@dsn.okisemi.com>
From: Toshiharu Okada <toshiharu-linux@dsn.okisemi.com>
Date: Fri, 2 Sep 2011 09:20:08 +0900
> This patch added the processing which should be done to hardware,
> when a FIFO over run error occurred.
>
> Signed-off-by: Toshiharu Okada <toshiharu-linux@dsn.okisemi.com>
Applied.
^ permalink raw reply
* Re: [PATCH 3/3] pch_gbe: support ML7831 IOH
From: David Miller @ 2011-09-15 21:26 UTC (permalink / raw)
To: toshiharu-linux
Cc: netdev, linux-kernel, eric.dumazet, qi.wang, yong.y.wang,
joel.clark, kok.howg.ewe, tomoya-linux
In-Reply-To: <1314922809-4041-3-git-send-email-toshiharu-linux@dsn.okisemi.com>
From: Toshiharu Okada <toshiharu-linux@dsn.okisemi.com>
Date: Fri, 2 Sep 2011 09:20:09 +0900
> Support new device OKI SEMICONDUCTOR ML7831 IOH(Input/Output Hub)
>
> ML7831 is for general purpose use.
> ML7831 is companion chip for Intel Atom E6xx series.
> ML7831 is completely compatible for Intel EG20T PCH.
>
> Signed-off-by: Toshiharu Okada <toshiharu-linux@dsn.okisemi.com>
Applied.
^ permalink raw reply
* Re: [patch net-next-2.6] net: consolidate and fix ethtool_ops->get_settings calling
From: David Miller @ 2011-09-15 21:33 UTC (permalink / raw)
To: jpirko
Cc: netdev, ralf, fubar, andy, kaber, bprakash, JBottomley,
robert.w.love, shemminger, decot, bhutchings, mirq-linux,
alexander.h.duyck, amit.salecha, eric.dumazet, therbert, paulmck,
laijs, xiaosuo, greearb, loke.chetan, linux-mips, linux-scsi,
devel, bridge
In-Reply-To: <1314905304-16485-1-git-send-email-jpirko@redhat.com>
From: Jiri Pirko <jpirko@redhat.com>
Date: Thu, 1 Sep 2011 21:28:24 +0200
> This patch does several things:
> - introduces __ethtool_get_settings which is called from ethtool code and
> from dev_ethtool_get_settings() as well.
> - dev_ethtool_get_settings() becomes rtnl wrapper for
> __ethtool_get_settings()
> - changes calling in drivers so rtnl locking is respected. In
> iboe_get_rate was previously ->get_settings() called unlocked. This
> fixes it
> - introduces rtnl_lock in bnx2fc_vport_create() and fcoe_vport_create()
> so bnx2fc_if_create() and fcoe_if_create() are called locked as they
> are from other places.
> - prb_calc_retire_blk_tmo() in af_packet.c was not calling get_settings
> with rtnl_lock. So use dev_ethtool_get_settings here.
> - use __ethtool_get_settings() in bonding code
>
> Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Applied, thanks.
^ permalink raw reply
* macvtap mode private - IFF_PROMISC flag visibility for userspace
From: Christian Benvenuti (benve) @ 2011-09-15 21:56 UTC (permalink / raw)
To: netdev; +Cc: mst, Roopa Prabhu (roprabhu), David Wang (dwang2)
When you use/create a macvtap interface in passthrou mode, the lowerdev
is put
into promiscuous mode (NOTE: this may change if/once Roopa's rx_filter
changes
get merged - http://www.spinics.net/lists/netdev/msg174245.html).
However, macvtap also puts the lowerdev into promiscuous mode when
- mode is "macvtap private"
AND
- the lowerdev does not support ndo_set_rx_mode
This would be the codepath in the latter case:
macvlan_open
|
+--> dev_uc_add
|
+--> __dev_set_rx_mode
|
+--> if (!ndo_set_rx_mode)
__dev_set_promiscuity
The difference between the two cases above (passthru vs private) however
is that the status of IFF_PROMISC is not exported the same way to user
space:
In the second case (private) dev_get_flags does not return the
IFF_PROMISC flag.
This means that, for example, neither 'ifconfig' nor 'ip link' show that
the
lowerdev is in promisc mode. This does not help throubleshooting.
Yes, the kernel generates a log message when it changes the promiscuity
mode
of the card, but...
Yes, you can check whether IFF_PROMISC is set or not by looking at
/sys/class/net/<dev_name>/flags
but I can't see why dev_get_flags should not return it too.
Right now dev_get_flags returns IFF_PROMISC only if it set in
dev->gflags, ie it
ignores dev->flags & IFF_PROMISC.
Is this the expected behavior by design?
/Chris
^ permalink raw reply
* Very confused about broute DROP
From: Greg Scott @ 2011-09-15 22:48 UTC (permalink / raw)
To: netdev; +Cc: Graham Parenteau
In-Reply-To: <925A849792280C4E80C5461017A4B8A2A0414B@mail733.InfraSupportEtc.com>
I don't get this. Why does:
ebtables -t broute -A BROUTING -j DROP
completely knock a Linux host offline?
This is what the man page for ebtables says:
The targets DROP and ACCEPT have a special meaning in the broute table
(these names are used instead of more descriptive names to keep the
implementation generic). DROP actually means the frame has to be
routed, while ACCEPT means the frame has to be bridged. The BROUTING
chain is traversed very early. However, it is only traversed by
frames entering on a bridge port that is in forwarding state.
Normally those frames would be bridged, but you can decide otherwise
here. The redirect target is very handy here.
So based on the above paragraph, I should be able to do something like
this:
# Here is what to bridge
ebtables -t broute -A BROUTING -p IPv4 --ip-destination $PUBLIC_IP1 -j
ACCEPT
ebtables -t broute -A BROUTING -p IPv4 --ip-destination $PUBLIC_IP2 -j
ACCEPT
# Route everything else
ebtables -t broute -A BROUTING -j DROP
So I tried above and knocked that box completely offline. I'm missing
something.
Here is what the paragraph about redirect in the ebtables man pages
says:
The redirect target will change the MAC target address to that of the
bridge device the frame arrived on. This target can only be used in the
BROUTING chain of the broute table and the PREROUTING chain of the nat
table. In the BROUTING chain, the MAC address of the bridge port is
used as destination address, in the PREROUTING chain, the MAC address of
the bridge is used.
OK - so this target MAC address - is this the MAC Address of an ethnn
port that's part of the bridge, or the MAC Address of another node? I
was thinking it was the MAC Address of another node, but maybe it's just
the MAC Address of a port on this bridge?
And there are some examples here:
http://ebtables.sourceforge.net/examples/basic.html#ex_redirect
that I really don't get. So instead of trial and error guessing, I
figured I would ask.
If anyone can help me understand this, I'll take a stab at writing it up
as clearly as I know how for use in future versions of man pages.
Thanks
- Greg Scott
^ permalink raw reply
* RE: Very confused about broute DROP
From: Christian Benvenuti (benve) @ 2011-09-15 23:08 UTC (permalink / raw)
To: Greg Scott, netdev; +Cc: Graham Parenteau
In-Reply-To: <925A849792280C4E80C5461017A4B8A2A04438@mail733.InfraSupportEtc.com>
How about ARP? You need it too ...
> -----Original Message-----
> From: netdev-owner@vger.kernel.org [mailto:netdev-
> owner@vger.kernel.org] On Behalf Of Greg Scott
> Sent: Thursday, September 15, 2011 3:48 PM
> To: netdev@vger.kernel.org
> Cc: Graham Parenteau
> Subject: Very confused about broute DROP
>
> I don't get this. Why does:
>
> ebtables -t broute -A BROUTING -j DROP
>
> completely knock a Linux host offline?
>
> This is what the man page for ebtables says:
>
> The targets DROP and ACCEPT have a special meaning in the broute table
> (these names are used instead of more descriptive names to keep the
> implementation generic). DROP actually means the frame has to be
> routed, while ACCEPT means the frame has to be bridged. The BROUTING
> chain is traversed very early. However, it is only traversed by
> frames entering on a bridge port that is in forwarding state.
> Normally those frames would be bridged, but you can decide otherwise
> here. The redirect target is very handy here.
>
> So based on the above paragraph, I should be able to do something like
> this:
>
> # Here is what to bridge
> ebtables -t broute -A BROUTING -p IPv4 --ip-destination $PUBLIC_IP1 -j
> ACCEPT
> ebtables -t broute -A BROUTING -p IPv4 --ip-destination $PUBLIC_IP2 -j
> ACCEPT
>
> # Route everything else
> ebtables -t broute -A BROUTING -j DROP
>
> So I tried above and knocked that box completely offline. I'm missing
> something.
>
> Here is what the paragraph about redirect in the ebtables man pages
> says:
>
> The redirect target will change the MAC target address to that of the
> bridge device the frame arrived on. This target can only be used in
the
> BROUTING chain of the broute table and the PREROUTING chain of the
> nat
> table. In the BROUTING chain, the MAC address of the bridge port
> is
> used as destination address, in the PREROUTING chain, the MAC address
> of
> the bridge is used.
>
> OK - so this target MAC address - is this the MAC Address of an ethnn
> port that's part of the bridge, or the MAC Address of another node? I
> was thinking it was the MAC Address of another node, but maybe it's
> just
> the MAC Address of a port on this bridge?
>
> And there are some examples here:
> http://ebtables.sourceforge.net/examples/basic.html#ex_redirect
>
> that I really don't get. So instead of trial and error guessing, I
> figured I would ask.
>
> If anyone can help me understand this, I'll take a stab at writing it
> up
> as clearly as I know how for use in future versions of man pages.
>
> Thanks
>
> - Greg Scott
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* 群发软件+买家搜索机+109届广交会买家、海关数据,B2B询盘买家500万。
From: 仅10元每天 @ 2011-09-16 0:22 UTC (permalink / raw)
群发软件+109届广交会买家、海关数据、搜索引擎买家,B2B询盘买家共500万,仅10元每天。
保证每天都有买家回复。
保证每天都有买家回复。
1、群发软件： 操作简单，功能强大，模仿人工操作模式，到达率高，日发送5万封以上。
2、500万买家资源： 赠送的500万买家资源库，每月更新 。
3、超级海外买家Email搜索机： 每天能搜索1-2万以上买家真实EMAIL，成单率高。
要的抓紧联系QQ: 1339625218 或者立即回复邮箱: 1339625218@qq.com
要的抓紧联系QQ: 1339625218 或者立即回复邮箱: 1339625218@qq.com
要的抓紧联系QQ: 1339625218 或者立即回复邮箱: 1339625218@qq.com
免费赠送:
一共8个包(数据是全行业的，按照行业分好类，并且可以按照关键词查询的)：
1，2011春季109届广交会买家现场询盘数据库新鲜出炉，超级新鲜买家，新鲜数据，容易成单！
2，购买后可以免费更新2011秋季广交会+2012春季广交会买家数据。太超值了。
3，最新全球买家库,共451660条数据。 (最新更新日期 2011-05-16日)
4，2008年,2009年,2010年 春季+秋季广交会买家名录，103 104 105 106 107 108 共六届 共120.6万数据。
5，2010年国际促销协会（PPAI）成员名单 PPAI Members Directory，非常重要的大买家。
6，2010年到香港采购的国外客人名录(香港贸发局提供)，共7.2万数据，超级重要的买家。
7，48.68万条最新买家询盘，购买后每月更新 1-2万条，包括2部分，1，最新的询盘 2，最新的展会买家。免费更新36个月。
8，2009年海关提单数据piers版数据 1千万。
诚信为本，支持支付宝担保交易 (先发货并安装设置群发软件，然后付款) 彻底打消您的 顾虑。
精准数据-成单率极高
精准数据-成单率极高
精准数据-成单率极高
精准数据-成单率极高
精准数据-成单率极高
^ permalink raw reply
* Re: [PATCH net-next-2.6] e1000: don't enable dma receives until after dma address has been setup
From: Andy Gospodarek @ 2011-09-16 1:50 UTC (permalink / raw)
To: Jesse Brandeburg
Cc: Michael S. Tsirkin, e1000-devel, netdev@vger.kernel.org,
Dean Nelson, Andy Gospodarek
In-Reply-To: <20110915102112.0000146b@unknown>
On Thu, Sep 15, 2011 at 10:21:12AM -0700, Jesse Brandeburg wrote:
> On Wed, 14 Sep 2011 17:31:38 -0700
> Dean Nelson <dnelson@redhat.com> wrote:
>
> > Doing an 'ifconfig ethN down' followed by an 'ifconfig ethN up' on a
> > qemu-kvm guest system configured with two e1000 NICs can result in an
> > 'unable to handle kernel paging request at 0000000100000000' or 'bad
> > page map in process ...' or something similar.
>
> <snip>
>
> > The corruption appears to result from the following...
> >
> > . An 'ifconfig ethN down' gets us into e1000_close(), which through
> > a number of subfunctions results in:
> > 1. E1000_RCTL_EN being cleared in RCTL register. [e1000_down()]
> > 2. dma_free_coherent() being called. [e1000_free_rx_resources()]
> >
> > . An 'ifconfig ethN up' gets us into e1000_open(), which through a
> > number of subfunctions results in:
> > 1. dma_alloc_coherent() being called.
> > [e1000_setup_rx_resources()] 2. E1000_RCTL_EN being set in RCTL
> > register. [e1000_setup_rctl()] 3. E1000_RCTL_EN being cleared in
> > RCTL register. [e1000_configure_rx()] 4. RDLEN, RDBAH and RDBAL
> > registers being set to reflect the dma page allocated in step 1.
> > [e1000_configure_rx()] 5. E1000_RCTL_EN being set in RCTL register.
> > [e1000_configure_rx()]
> >
> > During the 'ifconfig ethN up' there is a window opened, starting in
> > step 2 where the receives are enabled up until they are disabled in
> > step 3, in which the address of the receive descriptor dma page known
> > by the NIC is still the previous one which was freed during the
> > 'ifconfig ethN down'. If this memory has been reallocated for some
> > other use and the NIC feels so inclined, it will write to that former
> > dma page with predictably unpleasant results.
> >
> > I realize that in the guest, we're dealing with an e1000 NIC that is
> > software emulated by qemu-kvm. The problem doesn't appear to occur on
> > bare-metal. Andy suspects that this is because in the emulator
> > link-up is essentially instant and traffic can start flowing
> > immediately. Whereas on bare-metal, link-up usually seems to take at
> > least a few milliseconds. And this might be enough to prevent traffic
> > from flowing into the device inside the window where E1000_RCTL_EN is
> > set.
>
> nice analysis dean, yes, we shouldn't enable rx before we have the
> hardware all ready.
>
> You didn't mention however that the hardware is reset in e1000_down,
> which will clear the RDBAL/RDBAH in real hardware.
>
> >
> > So perhaps a modification needs to be made to the qemu-kvm e1000 NIC
> > emulator to delay the link-up. But in defense of the emulator, it
> > seems like a bad idea to enable dma operations before the address of
> > the memory to be involved has been made known.
>
> the hardware reset code in kvm should also reset to default many
> registers (almost all of them in fact) which may also end up solving
> the problem.
>
> >
> > The following patch no longer enables receives in e1000_setup_rctl()
> > but leaves them however they were. It only enables receives in
> > e1000_configure_rx(), and only after the dma address has been made
> > known to the hardware.
>
> I still like your patch better as it is more correct. We could also
> correct the kvm virtual hardware driver.
>
I agree that the virtual hardware drivers should be fixed. I took a
quick look at the emulation code and despite the fact that there is some
reset code that clears out registers and sets them back to default
values, I don't see it getting called when the E1000_CTRL_RST is set in
the CTRL register.
The patch below might do the trick. This is totally untested, but it
seems appropriate based on my quick audit of that code and the driver's
expectations.
diff --git a/hw/e1000.c b/hw/e1000.c
index a6d12c5..e74dbf3 100644
--- a/hw/e1000.c
+++ b/hw/e1000.c
@@ -192,9 +192,14 @@ rxbufsize(uint32_t v)
return 2048;
}
+static void e1000_reset(void *opaque);
+
static void
set_ctrl(E1000State *s, int index, uint32_t val)
{
+ /* reset the hardware registers */
+ if (val & E1000_CTRL_RST)
+ e1000_reset(s);
/* RST is self clearing */
s->mac_reg[CTRL] = val & ~E1000_CTRL_RST;
}
------------------------------------------------------------------------------
BlackBerry® DevCon Americas, Oct. 18-20, San Francisco, CA
http://p.sf.net/sfu/rim-devcon-copy2
_______________________________________________
E1000-devel mailing list
E1000-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel® Ethernet, visit http://communities.intel.com/community/wired
^ permalink raw reply related
* Re: [PATCH net-next-2.6] e1000: don't enable dma receives until after dma address has been setup
From: Jeff Kirsher @ 2011-09-16 2:03 UTC (permalink / raw)
To: Dean Nelson
Cc: Brandeburg, Jesse, netdev@vger.kernel.org, Michael S. Tsirkin,
Andy Gospodarek, Dave, Tushar N,
e1000-devel@lists.sourceforge.net
In-Reply-To: <4E724267.2020205@redhat.com>
[-- Attachment #1: Type: text/plain, Size: 3425 bytes --]
On Thu, 2011-09-15 at 11:22 -0700, Dean Nelson wrote:
> On 09/15/2011 12:21 PM, Jesse Brandeburg wrote:
> > On Wed, 14 Sep 2011 17:31:38 -0700
> > Dean Nelson<dnelson@redhat.com> wrote:
> >
> >> Doing an 'ifconfig ethN down' followed by an 'ifconfig ethN up' on a
> >> qemu-kvm guest system configured with two e1000 NICs can result in an
> >> 'unable to handle kernel paging request at 0000000100000000' or 'bad
> >> page map in process ...' or something similar.
> >
> > <snip>
> >
> >> The corruption appears to result from the following...
> >>
> <snip>
> >>
> >> I realize that in the guest, we're dealing with an e1000 NIC that is
> >> software emulated by qemu-kvm. The problem doesn't appear to occur on
> >> bare-metal. Andy suspects that this is because in the emulator
> >> link-up is essentially instant and traffic can start flowing
> >> immediately. Whereas on bare-metal, link-up usually seems to take at
> >> least a few milliseconds. And this might be enough to prevent traffic
> >> from flowing into the device inside the window where E1000_RCTL_EN is
> >> set.
> >
> > nice analysis dean, yes, we shouldn't enable rx before we have the
> > hardware all ready.
>
> Thank you.
>
>
> > You didn't mention however that the hardware is reset in e1000_down,
> > which will clear the RDBAL/RDBAH in real hardware.
>
> You are correct, I did fail to mention the reset. And the clearing of
> RDBAL/RDHAH was definitely not happening in the qemu-kvm emulator.
>
>
> >> So perhaps a modification needs to be made to the qemu-kvm e1000 NIC
> >> emulator to delay the link-up. But in defense of the emulator, it
> >> seems like a bad idea to enable dma operations before the address of
> >> the memory to be involved has been made known.
> >
> > the hardware reset code in kvm should also reset to default many
> > registers (almost all of them in fact) which may also end up solving
> > the problem.
>
> Agreed.
>
>
> >> The following patch no longer enables receives in e1000_setup_rctl()
> >> but leaves them however they were. It only enables receives in
> >> e1000_configure_rx(), and only after the dma address has been made
> >> known to the hardware.
> >
> > I still like your patch better as it is more correct. We could also
> > correct the kvm virtual hardware driver.
>
> The hardware emulator should definitely be doing a proper hardware reset.
>
>
> >> There are two places where e1000_setup_rctl() gets called. The one in
> <snip>
> >>
> >> The e1000e looks to have the same issue. I don't know about igb. But
> >> I'm not aware of either having hardware emulation in qemu-kvm. So
> >> unless this issue is reproducible on bare-metal... it's probably not
> >> a big deal for them.
> >>
> <snip>
> >
> > generally i like the patch. We should take it in and test it, and I
> > don't really see any problems with it.
>
> Thanks.
>
> As mentioned above, the e1000e has a similar algorithm, but the
> FLAG2_NO_DISABLE_RX complicates it a bit. I have no idea what happens
> if receives are enabled while setting RDBAL and RDBAH. Is there any
> possibility that the hardware could try to make use of a half-baked
> address?
>
> Thanks much for your review of the patch.
>
> Dean
>
Thanks Dean! I will add your patch to my queue. I will work with Bruce
to review e1000e and get a patch put together.
[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 490 bytes --]
^ permalink raw reply
* Re: net GIT trees on github
From: Jeff Kirsher @ 2011-09-16 3:01 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-wireless, netfilter-devel
In-Reply-To: <20110915.150552.93591736515744753.davem@davemloft.net>
On Thu, Sep 15, 2011 at 12:05, David Miller <davem@davemloft.net> wrote:
>
> I finally broke down and started putting some trees up on github.
>
> The plain 'net' tree is up already:
>
> git://github.com/davem330/net.git
>
> And a "net-next.git" will be there shortly.
>
> This is a temporary situation so that I can push out the backlog
> a little bit until kernel.org is fully back in service. Once
> kernel.org is back I will stop updating these trees.
>
I have forked your net-next tree for Intel Wired LAN Driver updates,
here is the link
git://github.com/Jkirsher/net-next.git
My pull requests will use this tree (only while master.kernel.org is
down). This is meant as a FYI and will still send out a cover page
with this info as well.
--
Cheers,
Jeff
^ permalink raw reply
* RE: Very confused about broute DROP
From: Greg Scott @ 2011-09-16 3:19 UTC (permalink / raw)
To: Christian Benvenuti (benve), netdev; +Cc: Graham Parenteau
In-Reply-To: <184D23435BECB444AB6B9D4630C8EC83028548FD@XMB-RCD-303.cisco.com>
If I'm reading the ebtables ARP stuff right, it looks like I can use
this for selective proxy ARPs. This may have made my life much nicer
with a project that didn't work out very well 5 years ago when I
inherited a network with systems that had public IP Addresses on both
sides of the firewall. For now, why did
ebtables -t broute -A BROUTING -j DROP
shut my system down and force me to drive over there at 4AM this morning
to undo it? It's not supposed to really drop, it's supposed to route
it. And those examples aren't making any sense to me.
Thanks
- Greg
-----Original Message-----
From: Christian Benvenuti (benve) [mailto:benve@cisco.com]
Sent: Thursday, September 15, 2011 6:08 PM
To: Greg Scott; netdev@vger.kernel.org
Cc: Graham Parenteau
Subject: RE: Very confused about broute DROP
How about ARP? You need it too ...
^ permalink raw reply
* RE: Very confused about broute DROP
From: Christian Benvenuti (benve) @ 2011-09-16 4:23 UTC (permalink / raw)
To: Greg Scott, netdev; +Cc: Graham Parenteau
In-Reply-To: <925A849792280C4E80C5461017A4B8A2A0443A@mail733.InfraSupportEtc.com>
What I meant is that your host needs to be able to route
(which means ... to process) its own ARP traffic, ... IPv4
does not work without ARP, right?
This means you need to add one more DROP rule for the ARP
traffic that is addressed to the MAC of the host interfaces
(nothing to do with proxy ARP).
/Chris
> -----Original Message-----
> From: Greg Scott [mailto:GregScott@Infrasupport.com]
> Sent: Thursday, September 15, 2011 8:19 PM
> To: Christian Benvenuti (benve); netdev@vger.kernel.org
> Cc: Graham Parenteau
> Subject: RE: Very confused about broute DROP
>
> If I'm reading the ebtables ARP stuff right, it looks like I can use
> this for selective proxy ARPs. This may have made my life much nicer
> with a project that didn't work out very well 5 years ago when I
> inherited a network with systems that had public IP Addresses on both
> sides of the firewall. For now, why did
>
> ebtables -t broute -A BROUTING -j DROP
>
> shut my system down and force me to drive over there at 4AM this
> morning
> to undo it? It's not supposed to really drop, it's supposed to route
> it. And those examples aren't making any sense to me.
>
> Thanks
>
> - Greg
>
>
>
> -----Original Message-----
> From: Christian Benvenuti (benve) [mailto:benve@cisco.com]
> Sent: Thursday, September 15, 2011 6:08 PM
> To: Greg Scott; netdev@vger.kernel.org
> Cc: Graham Parenteau
> Subject: RE: Very confused about broute DROP
>
> How about ARP? You need it too ...
>
^ permalink raw reply
* [net-next 00/11][pull request] Intel Wired LAN Driver Updates
From: Jeff Kirsher @ 2011-09-16 4:42 UTC (permalink / raw)
To: davem; +Cc: Jeff Kirsher, netdev, gospo
The following series contains updates to ixgbe only. These are primarily
cleanups of the ixgbe driver. The first two patches of the series:
ixgbe: Change default Tx work limit size to 256 buffers
ixgbe: consolidate all MSI-X ring interrupts and poll routines into one
are re-worked based on previous community feedback (Dave and Ben).
The following are changes since commit 4bc71cb983fd2844e603bf633df2bb53385182d2:
net: consolidate and fix ethtool_ops->get_settings calling
and are available in the git repository at:
git://github.com/Jkirsher/net-next.git
Alexander Duyck (11):
ixgbe: Change default Tx work limit size to 256 buffers
v2 ixgbe: consolidate all MSI-X ring interrupts and poll routines
into one
ixgbe: cleanup allocation and freeing of IRQ affinity hint
ixgbe: Use ring->dev instead of adapter->pdev->dev when updating DCA
ixgbe: commonize ixgbe_map_rings_to_vectors to work for all interrupt
types
ixgbe: Drop unnecessary adapter->hw dereference in loopback test
setup
ixgbe: combine PCI_VDEVICE and board declaration to same line
ixgbe: Update TXDCTL configuration to correctly handle WTHRESH
ixgbe: cleanup reset paths
ixgbe: cleanup configuration of EITRSEL and VF reset path
ixgbe: Correctly name and handle MSI-X other interrupt
drivers/net/ethernet/intel/ixgbe/ixgbe.h | 2 +-
drivers/net/ethernet/intel/ixgbe/ixgbe_82598.c | 13 +-
drivers/net/ethernet/intel/ixgbe/ixgbe_82599.c | 40 +-
drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c | 18 +-
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 775 ++++++++--------------
drivers/net/ethernet/intel/ixgbe/ixgbe_type.h | 1 +
drivers/net/ethernet/intel/ixgbe/ixgbe_x540.c | 72 +--
7 files changed, 341 insertions(+), 580 deletions(-)
--
1.7.6
^ permalink raw reply
* [net-next 02/11] v2 ixgbe: consolidate all MSI-X ring interrupts and poll routines into one
From: Jeff Kirsher @ 2011-09-16 4:42 UTC (permalink / raw)
To: davem; +Cc: Alexander Duyck, netdev, gospo, Jeff Kirsher
In-Reply-To: <1316148173-11209-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Alexander Duyck <alexander.h.duyck@intel.com>
This change consolidates all of the MSI-X interrupt and polling routines
into two single functions. One for the interrupt and one for the code.
The main advantage to doing this is that the compiler can optimize the
routines into single monolithic functions which should allow all of them
function to occupy a single block of memory and as such avoid jumping
around.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 279 ++++++-------------------
1 files changed, 63 insertions(+), 216 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index 0f633ad..3ce0277 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -1297,9 +1297,9 @@ static inline bool ixgbe_get_rsc_state(union ixgbe_adv_rx_desc *rx_desc)
IXGBE_RXDADV_RSCCNT_MASK);
}
-static void ixgbe_clean_rx_irq(struct ixgbe_q_vector *q_vector,
+static bool ixgbe_clean_rx_irq(struct ixgbe_q_vector *q_vector,
struct ixgbe_ring *rx_ring,
- int *work_done, int work_to_do)
+ int budget)
{
struct ixgbe_adapter *adapter = q_vector->adapter;
union ixgbe_adv_rx_desc *rx_desc, *next_rxd;
@@ -1479,11 +1479,11 @@ static void ixgbe_clean_rx_irq(struct ixgbe_q_vector *q_vector,
#endif /* IXGBE_FCOE */
ixgbe_receive_skb(q_vector, skb, staterr, rx_ring, rx_desc);
+ budget--;
next_desc:
rx_desc->wb.upper.status_error = 0;
- (*work_done)++;
- if (*work_done >= work_to_do)
+ if (!budget)
break;
/* return some buffers to hardware, one at a time is too slow */
@@ -1524,9 +1524,10 @@ next_desc:
u64_stats_update_end(&rx_ring->syncp);
q_vector->rx.total_packets += total_rx_packets;
q_vector->rx.total_bytes += total_rx_bytes;
+
+ return !!budget;
}
-static int ixgbe_clean_rxonly(struct napi_struct *, int);
/**
* ixgbe_configure_msix - Configure MSI-X hardware
* @adapter: board private structure
@@ -1980,167 +1981,18 @@ static inline void ixgbe_irq_disable_queues(struct ixgbe_adapter *adapter,
/* skip the flush */
}
-static irqreturn_t ixgbe_msix_clean_tx(int irq, void *data)
-{
- struct ixgbe_q_vector *q_vector = data;
-
- if (!q_vector->tx.count)
- return IRQ_HANDLED;
-
- /* EIAM disabled interrupts (on this vector) for us */
- napi_schedule(&q_vector->napi);
-
- return IRQ_HANDLED;
-}
-
-/**
- * ixgbe_msix_clean_rx - single unshared vector rx clean (all queues)
- * @irq: unused
- * @data: pointer to our q_vector struct for this interrupt vector
- **/
-static irqreturn_t ixgbe_msix_clean_rx(int irq, void *data)
+static irqreturn_t ixgbe_msix_clean_rings(int irq, void *data)
{
struct ixgbe_q_vector *q_vector = data;
- if (!q_vector->rx.count)
- return IRQ_HANDLED;
-
/* EIAM disabled interrupts (on this vector) for us */
- napi_schedule(&q_vector->napi);
-
- return IRQ_HANDLED;
-}
-
-static irqreturn_t ixgbe_msix_clean_many(int irq, void *data)
-{
- struct ixgbe_q_vector *q_vector = data;
- if (!q_vector->tx.count && !q_vector->rx.count)
- return IRQ_HANDLED;
-
- /* EIAM disabled interrupts (on this vector) for us */
- napi_schedule(&q_vector->napi);
+ if (q_vector->rx.ring || q_vector->tx.ring)
+ napi_schedule(&q_vector->napi);
return IRQ_HANDLED;
}
-/**
- * ixgbe_clean_rxonly - msix (aka one shot) rx clean routine
- * @napi: napi struct with our devices info in it
- * @budget: amount of work driver is allowed to do this pass, in packets
- *
- * This function is optimized for cleaning one queue only on a single
- * q_vector!!!
- **/
-static int ixgbe_clean_rxonly(struct napi_struct *napi, int budget)
-{
- struct ixgbe_q_vector *q_vector =
- container_of(napi, struct ixgbe_q_vector, napi);
- struct ixgbe_adapter *adapter = q_vector->adapter;
- int work_done = 0;
-
-#ifdef CONFIG_IXGBE_DCA
- if (adapter->flags & IXGBE_FLAG_DCA_ENABLED)
- ixgbe_update_dca(q_vector);
-#endif
-
- ixgbe_clean_rx_irq(q_vector, q_vector->rx.ring, &work_done, budget);
-
- /* If all Rx work done, exit the polling mode */
- if (work_done < budget) {
- napi_complete(napi);
- if (adapter->rx_itr_setting & 1)
- ixgbe_set_itr(q_vector);
- if (!test_bit(__IXGBE_DOWN, &adapter->state))
- ixgbe_irq_enable_queues(adapter,
- ((u64)1 << q_vector->v_idx));
- }
-
- return work_done;
-}
-
-/**
- * ixgbe_clean_rxtx_many - msix (aka one shot) rx clean routine
- * @napi: napi struct with our devices info in it
- * @budget: amount of work driver is allowed to do this pass, in packets
- *
- * This function will clean more than one rx queue associated with a
- * q_vector.
- **/
-static int ixgbe_clean_rxtx_many(struct napi_struct *napi, int budget)
-{
- struct ixgbe_q_vector *q_vector =
- container_of(napi, struct ixgbe_q_vector, napi);
- struct ixgbe_adapter *adapter = q_vector->adapter;
- struct ixgbe_ring *ring;
- int work_done = 0;
- bool clean_complete = true;
-
-#ifdef CONFIG_IXGBE_DCA
- if (adapter->flags & IXGBE_FLAG_DCA_ENABLED)
- ixgbe_update_dca(q_vector);
-#endif
-
- for (ring = q_vector->tx.ring; ring != NULL; ring = ring->next)
- clean_complete &= ixgbe_clean_tx_irq(q_vector, ring);
-
- /* attempt to distribute budget to each queue fairly, but don't allow
- * the budget to go below 1 because we'll exit polling */
- budget /= (q_vector->rx.count ?: 1);
- budget = max(budget, 1);
-
- for (ring = q_vector->rx.ring; ring != NULL; ring = ring->next)
- ixgbe_clean_rx_irq(q_vector, ring, &work_done, budget);
-
- if (!clean_complete)
- work_done = budget;
-
- /* If all Rx work done, exit the polling mode */
- if (work_done < budget) {
- napi_complete(napi);
- if (adapter->rx_itr_setting & 1)
- ixgbe_set_itr(q_vector);
- if (!test_bit(__IXGBE_DOWN, &adapter->state))
- ixgbe_irq_enable_queues(adapter,
- ((u64)1 << q_vector->v_idx));
- return 0;
- }
-
- return work_done;
-}
-
-/**
- * ixgbe_clean_txonly - msix (aka one shot) tx clean routine
- * @napi: napi struct with our devices info in it
- * @budget: amount of work driver is allowed to do this pass, in packets
- *
- * This function is optimized for cleaning one queue only on a single
- * q_vector!!!
- **/
-static int ixgbe_clean_txonly(struct napi_struct *napi, int budget)
-{
- struct ixgbe_q_vector *q_vector =
- container_of(napi, struct ixgbe_q_vector, napi);
- struct ixgbe_adapter *adapter = q_vector->adapter;
-
-#ifdef CONFIG_IXGBE_DCA
- if (adapter->flags & IXGBE_FLAG_DCA_ENABLED)
- ixgbe_update_dca(q_vector);
-#endif
-
- if (!ixgbe_clean_tx_irq(q_vector, q_vector->tx.ring))
- return budget;
-
- /* If all Tx work done, exit the polling mode */
- napi_complete(napi);
- if (adapter->tx_itr_setting & 1)
- ixgbe_set_itr(q_vector);
- if (!test_bit(__IXGBE_DOWN, &adapter->state))
- ixgbe_irq_enable_queues(adapter, ((u64)1 << q_vector->v_idx));
-
- return 0;
-}
-
static inline void map_vector_to_rxq(struct ixgbe_adapter *a, int v_idx,
int r_idx)
{
@@ -2241,7 +2093,6 @@ out:
static int ixgbe_request_msix_irqs(struct ixgbe_adapter *adapter)
{
struct net_device *netdev = adapter->netdev;
- irqreturn_t (*handler)(int, void *);
int i, vector, q_vectors, err;
int ri = 0, ti = 0;
@@ -2252,31 +2103,25 @@ static int ixgbe_request_msix_irqs(struct ixgbe_adapter *adapter)
if (err)
return err;
-#define SET_HANDLER(_v) (((_v)->rx.count && (_v)->tx.count) \
- ? &ixgbe_msix_clean_many : \
- (_v)->rx.count ? &ixgbe_msix_clean_rx : \
- (_v)->tx.count ? &ixgbe_msix_clean_tx : \
- NULL)
for (vector = 0; vector < q_vectors; vector++) {
struct ixgbe_q_vector *q_vector = adapter->q_vector[vector];
- handler = SET_HANDLER(q_vector);
- if (handler == &ixgbe_msix_clean_rx) {
+ if (q_vector->tx.ring && q_vector->rx.ring) {
snprintf(q_vector->name, sizeof(q_vector->name) - 1,
- "%s-%s-%d", netdev->name, "rx", ri++);
- } else if (handler == &ixgbe_msix_clean_tx) {
+ "%s-%s-%d", netdev->name, "TxRx", ri++);
+ ti++;
+ } else if (q_vector->rx.ring) {
snprintf(q_vector->name, sizeof(q_vector->name) - 1,
- "%s-%s-%d", netdev->name, "tx", ti++);
- } else if (handler == &ixgbe_msix_clean_many) {
+ "%s-%s-%d", netdev->name, "rx", ri++);
+ } else if (q_vector->tx.ring) {
snprintf(q_vector->name, sizeof(q_vector->name) - 1,
- "%s-%s-%d", netdev->name, "TxRx", ri++);
- ti++;
+ "%s-%s-%d", netdev->name, "tx", ti++);
} else {
/* skip this unused q_vector */
continue;
}
err = request_irq(adapter->msix_entries[vector].vector,
- handler, 0, q_vector->name,
+ &ixgbe_msix_clean_rings, 0, q_vector->name,
q_vector);
if (err) {
e_err(probe, "request_irq failed for MSIX interrupt "
@@ -2484,8 +2329,8 @@ static void ixgbe_free_irq(struct ixgbe_adapter *adapter)
i--;
for (; i >= 0; i--) {
/* free only the irqs that were actually requested */
- if (!adapter->q_vector[i]->rx.count &&
- !adapter->q_vector[i]->tx.count)
+ if (!adapter->q_vector[i]->rx.ring &&
+ !adapter->q_vector[i]->tx.ring)
continue;
free_irq(adapter->msix_entries[i].vector,
@@ -3478,19 +3323,8 @@ static void ixgbe_napi_enable_all(struct ixgbe_adapter *adapter)
q_vectors = 1;
for (q_idx = 0; q_idx < q_vectors; q_idx++) {
- struct napi_struct *napi;
q_vector = adapter->q_vector[q_idx];
- napi = &q_vector->napi;
- if (adapter->flags & IXGBE_FLAG_MSIX_ENABLED) {
- if (!q_vector->rx.count || !q_vector->tx.count) {
- if (q_vector->tx.count == 1)
- napi->poll = &ixgbe_clean_txonly;
- else if (q_vector->rx.count == 1)
- napi->poll = &ixgbe_clean_rxonly;
- }
- }
-
- napi_enable(napi);
+ napi_enable(&q_vector->napi);
}
}
@@ -4148,28 +3982,41 @@ static int ixgbe_poll(struct napi_struct *napi, int budget)
struct ixgbe_q_vector *q_vector =
container_of(napi, struct ixgbe_q_vector, napi);
struct ixgbe_adapter *adapter = q_vector->adapter;
- int tx_clean_complete, work_done = 0;
+ struct ixgbe_ring *ring;
+ int per_ring_budget;
+ bool clean_complete = true;
#ifdef CONFIG_IXGBE_DCA
if (adapter->flags & IXGBE_FLAG_DCA_ENABLED)
ixgbe_update_dca(q_vector);
#endif
- tx_clean_complete = ixgbe_clean_tx_irq(q_vector, adapter->tx_ring[0]);
- ixgbe_clean_rx_irq(q_vector, adapter->rx_ring[0], &work_done, budget);
+ for (ring = q_vector->tx.ring; ring != NULL; ring = ring->next)
+ clean_complete &= !!ixgbe_clean_tx_irq(q_vector, ring);
- if (!tx_clean_complete)
- work_done = budget;
+ /* attempt to distribute budget to each queue fairly, but don't allow
+ * the budget to go below 1 because we'll exit polling */
+ if (q_vector->rx.count > 1)
+ per_ring_budget = max(budget/q_vector->rx.count, 1);
+ else
+ per_ring_budget = budget;
- /* If budget not fully consumed, exit the polling mode */
- if (work_done < budget) {
- napi_complete(napi);
- if (adapter->rx_itr_setting & 1)
- ixgbe_set_itr(q_vector);
- if (!test_bit(__IXGBE_DOWN, &adapter->state))
- ixgbe_irq_enable_queues(adapter, IXGBE_EIMS_RTX_QUEUE);
- }
- return work_done;
+ for (ring = q_vector->rx.ring; ring != NULL; ring = ring->next)
+ clean_complete &= ixgbe_clean_rx_irq(q_vector, ring,
+ per_ring_budget);
+
+ /* If all work not completed, return budget and keep polling */
+ if (!clean_complete)
+ return budget;
+
+ /* all work done, exit the polling mode */
+ napi_complete(napi);
+ if (adapter->rx_itr_setting & 1)
+ ixgbe_set_itr(q_vector);
+ if (!test_bit(__IXGBE_DOWN, &adapter->state))
+ ixgbe_irq_enable_queues(adapter, ((u64)1 << q_vector->v_idx));
+
+ return 0;
}
/**
@@ -4810,19 +4657,15 @@ out:
**/
static int ixgbe_alloc_q_vectors(struct ixgbe_adapter *adapter)
{
- int q_idx, num_q_vectors;
+ int v_idx, num_q_vectors;
struct ixgbe_q_vector *q_vector;
- int (*poll)(struct napi_struct *, int);
- if (adapter->flags & IXGBE_FLAG_MSIX_ENABLED) {
+ if (adapter->flags & IXGBE_FLAG_MSIX_ENABLED)
num_q_vectors = adapter->num_msix_vectors - NON_Q_VECTORS;
- poll = &ixgbe_clean_rxtx_many;
- } else {
+ else
num_q_vectors = 1;
- poll = &ixgbe_poll;
- }
- for (q_idx = 0; q_idx < num_q_vectors; q_idx++) {
+ for (v_idx = 0; v_idx < num_q_vectors; v_idx++) {
q_vector = kzalloc_node(sizeof(struct ixgbe_q_vector),
GFP_KERNEL, adapter->node);
if (!q_vector)
@@ -4830,25 +4673,29 @@ static int ixgbe_alloc_q_vectors(struct ixgbe_adapter *adapter)
GFP_KERNEL);
if (!q_vector)
goto err_out;
+
q_vector->adapter = adapter;
+ q_vector->v_idx = v_idx;
+
if (q_vector->tx.count && !q_vector->rx.count)
q_vector->eitr = adapter->tx_eitr_param;
else
q_vector->eitr = adapter->rx_eitr_param;
- q_vector->v_idx = q_idx;
- netif_napi_add(adapter->netdev, &q_vector->napi, (*poll), 64);
- adapter->q_vector[q_idx] = q_vector;
+
+ netif_napi_add(adapter->netdev, &q_vector->napi,
+ ixgbe_poll, 64);
+ adapter->q_vector[v_idx] = q_vector;
}
return 0;
err_out:
- while (q_idx) {
- q_idx--;
- q_vector = adapter->q_vector[q_idx];
+ while (v_idx) {
+ v_idx--;
+ q_vector = adapter->q_vector[v_idx];
netif_napi_del(&q_vector->napi);
kfree(q_vector);
- adapter->q_vector[q_idx] = NULL;
+ adapter->q_vector[v_idx] = NULL;
}
return -ENOMEM;
}
@@ -6960,7 +6807,7 @@ static void ixgbe_netpoll(struct net_device *netdev)
int num_q_vectors = adapter->num_msix_vectors - NON_Q_VECTORS;
for (i = 0; i < num_q_vectors; i++) {
struct ixgbe_q_vector *q_vector = adapter->q_vector[i];
- ixgbe_msix_clean_many(0, q_vector);
+ ixgbe_msix_clean_rings(0, q_vector);
}
} else {
ixgbe_intr(adapter->pdev->irq, netdev);
--
1.7.6
^ permalink raw reply related
* [net-next 01/11] ixgbe: Change default Tx work limit size to 256 buffers
From: Jeff Kirsher @ 2011-09-16 4:42 UTC (permalink / raw)
To: davem; +Cc: Alexander Duyck, netdev, gospo, Jeff Kirsher
In-Reply-To: <1316148173-11209-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Alexander Duyck <alexander.h.duyck@intel.com>
This change makes it so that the default Tx work limit is 256 buffers or
1/2 of an entire ring instead of a full ring size so that it is much more
likely that we will be able to actually reach the work limit value.
Previously with the value set to an entire ring it would not have been
possible for us to trigger an event due to the fact that the Tx work is
stopped at the point where we cannot place one more buffer on the ring and
it is not restarted until cleanup is complete.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe.h | 1 +
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 8 ++++----
2 files changed, 5 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe.h b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
index 58482fc..3f5a744 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe.h
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
@@ -53,6 +53,7 @@
/* TX/RX descriptor defines */
#define IXGBE_DEFAULT_TXD 512
+#define IXGBE_DEFAULT_TX_WORK 256
#define IXGBE_MAX_TXD 4096
#define IXGBE_MIN_TXD 64
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index a30f826..0f633ad 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -804,7 +804,7 @@ static bool ixgbe_clean_tx_irq(struct ixgbe_q_vector *q_vector,
struct ixgbe_tx_buffer *tx_buffer;
union ixgbe_adv_tx_desc *tx_desc;
unsigned int total_bytes = 0, total_packets = 0;
- u16 budget = q_vector->tx.work_limit;
+ unsigned int budget = q_vector->tx.work_limit;
u16 i = tx_ring->next_to_clean;
tx_buffer = &tx_ring->tx_buffer_info[i];
@@ -891,7 +891,7 @@ static bool ixgbe_clean_tx_irq(struct ixgbe_q_vector *q_vector,
ixgbe_tx_timeout_reset(adapter);
/* the adapter is about to reset, no point in enabling stuff */
- return budget;
+ return true;
}
#define TX_WAKE_THRESHOLD (DESC_NEEDED * 2)
@@ -908,7 +908,7 @@ static bool ixgbe_clean_tx_irq(struct ixgbe_q_vector *q_vector,
}
}
- return budget;
+ return !!budget;
}
#ifdef CONFIG_IXGBE_DCA
@@ -5091,7 +5091,7 @@ static int __devinit ixgbe_sw_init(struct ixgbe_adapter *adapter)
adapter->rx_ring_count = IXGBE_DEFAULT_RXD;
/* set default work limits */
- adapter->tx_work_limit = adapter->tx_ring_count;
+ adapter->tx_work_limit = IXGBE_DEFAULT_TX_WORK;
/* initialize eeprom parameters */
if (ixgbe_init_eeprom_params_generic(hw)) {
--
1.7.6
^ permalink raw reply related
* [net-next 05/11] ixgbe: commonize ixgbe_map_rings_to_vectors to work for all interrupt types
From: Jeff Kirsher @ 2011-09-16 4:42 UTC (permalink / raw)
To: davem; +Cc: Alexander Duyck, netdev, gospo, Jeff Kirsher
In-Reply-To: <1316148173-11209-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Alexander Duyck <alexander.h.duyck@intel.com>
This patch makes it so that the map_rings_to_vectors call will work with
all interrupt types. The advantage to this is that there will now be a
predictable mapping for all given interrupt types.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 90 ++++++++++---------------
1 files changed, 35 insertions(+), 55 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index 0564c65..dd1b57b 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -2014,59 +2014,41 @@ static inline void map_vector_to_txq(struct ixgbe_adapter *a, int v_idx,
* group the rings as "efficiently" as possible. You would add new
* mapping configurations in here.
**/
-static int ixgbe_map_rings_to_vectors(struct ixgbe_adapter *adapter)
+static void ixgbe_map_rings_to_vectors(struct ixgbe_adapter *adapter)
{
- int q_vectors;
+ int q_vectors = adapter->num_msix_vectors - NON_Q_VECTORS;
+ int rxr_remaining = adapter->num_rx_queues, rxr_idx = 0;
+ int txr_remaining = adapter->num_tx_queues, txr_idx = 0;
int v_start = 0;
- int rxr_idx = 0, txr_idx = 0;
- int rxr_remaining = adapter->num_rx_queues;
- int txr_remaining = adapter->num_tx_queues;
- int i, j;
- int rqpv, tqpv;
- int err = 0;
- /* No mapping required if MSI-X is disabled. */
+ /* only one q_vector if MSI-X is disabled. */
if (!(adapter->flags & IXGBE_FLAG_MSIX_ENABLED))
- goto out;
-
- q_vectors = adapter->num_msix_vectors - NON_Q_VECTORS;
+ q_vectors = 1;
/*
- * The ideal configuration...
- * We have enough vectors to map one per queue.
+ * If we don't have enough vectors for a 1-to-1 mapping, we'll have to
+ * group them so there are multiple queues per vector.
+ *
+ * Re-adjusting *qpv takes care of the remainder.
*/
- if (q_vectors == adapter->num_rx_queues + adapter->num_tx_queues) {
- for (; rxr_idx < rxr_remaining; v_start++, rxr_idx++)
+ for (; v_start < q_vectors && rxr_remaining; v_start++) {
+ int rqpv = DIV_ROUND_UP(rxr_remaining, q_vectors - v_start);
+ for (; rqpv; rqpv--, rxr_idx++, rxr_remaining--)
map_vector_to_rxq(adapter, v_start, rxr_idx);
-
- for (; txr_idx < txr_remaining; v_start++, txr_idx++)
- map_vector_to_txq(adapter, v_start, txr_idx);
-
- goto out;
}
/*
- * If we don't have enough vectors for a 1-to-1
- * mapping, we'll have to group them so there are
- * multiple queues per vector.
+ * If there are not enough q_vectors for each ring to have it's own
+ * vector then we must pair up Rx/Tx on a each vector
*/
- /* Re-adjusting *qpv takes care of the remainder. */
- for (i = v_start; i < q_vectors; i++) {
- rqpv = DIV_ROUND_UP(rxr_remaining, q_vectors - i);
- for (j = 0; j < rqpv; j++) {
- map_vector_to_rxq(adapter, i, rxr_idx);
- rxr_idx++;
- rxr_remaining--;
- }
- tqpv = DIV_ROUND_UP(txr_remaining, q_vectors - i);
- for (j = 0; j < tqpv; j++) {
- map_vector_to_txq(adapter, i, txr_idx);
- txr_idx++;
- txr_remaining--;
- }
+ if ((v_start + txr_remaining) > q_vectors)
+ v_start = 0;
+
+ for (; v_start < q_vectors && txr_remaining; v_start++) {
+ int tqpv = DIV_ROUND_UP(txr_remaining, q_vectors - v_start);
+ for (; tqpv; tqpv--, txr_idx++, txr_remaining--)
+ map_vector_to_txq(adapter, v_start, txr_idx);
}
-out:
- return err;
}
/**
@@ -2083,10 +2065,6 @@ static int ixgbe_request_msix_irqs(struct ixgbe_adapter *adapter)
int vector, err;
int ri = 0, ti = 0;
- err = ixgbe_map_rings_to_vectors(adapter);
- if (err)
- return err;
-
for (vector = 0; vector < q_vectors; vector++) {
struct ixgbe_q_vector *q_vector = adapter->q_vector[vector];
struct msix_entry *entry = &adapter->msix_entries[vector];
@@ -2294,19 +2272,25 @@ static int ixgbe_request_irq(struct ixgbe_adapter *adapter)
struct net_device *netdev = adapter->netdev;
int err;
- if (adapter->flags & IXGBE_FLAG_MSIX_ENABLED) {
+ /* map all of the rings to the q_vectors */
+ ixgbe_map_rings_to_vectors(adapter);
+
+ if (adapter->flags & IXGBE_FLAG_MSIX_ENABLED)
err = ixgbe_request_msix_irqs(adapter);
- } else if (adapter->flags & IXGBE_FLAG_MSI_ENABLED) {
+ else if (adapter->flags & IXGBE_FLAG_MSI_ENABLED)
err = request_irq(adapter->pdev->irq, ixgbe_intr, 0,
netdev->name, adapter);
- } else {
+ else
err = request_irq(adapter->pdev->irq, ixgbe_intr, IRQF_SHARED,
netdev->name, adapter);
- }
- if (err)
+ if (err) {
e_err(probe, "request_irq failed, Error %d\n", err);
+ /* place q_vectors and rings back into a known good state */
+ ixgbe_reset_q_vectors(adapter);
+ }
+
return err;
}
@@ -2316,11 +2300,10 @@ static void ixgbe_free_irq(struct ixgbe_adapter *adapter)
int i, q_vectors;
q_vectors = adapter->num_msix_vectors;
-
i = q_vectors - 1;
free_irq(adapter->msix_entries[i].vector, adapter);
-
i--;
+
for (; i >= 0; i--) {
/* free only the irqs that were actually requested */
if (!adapter->q_vector[i]->rx.ring &&
@@ -2387,9 +2370,6 @@ static void ixgbe_configure_msi_and_legacy(struct ixgbe_adapter *adapter)
ixgbe_set_ivar(adapter, 0, 0, 0);
ixgbe_set_ivar(adapter, 1, 0, 0);
- map_vector_to_rxq(adapter, 0, 0);
- map_vector_to_txq(adapter, 0, 0);
-
e_info(hw, "Legacy interrupt IVAR setup done\n");
}
--
1.7.6
^ permalink raw reply related
* [net-next 03/11] ixgbe: cleanup allocation and freeing of IRQ affinity hint
From: Jeff Kirsher @ 2011-09-16 4:42 UTC (permalink / raw)
To: davem; +Cc: Alexander Duyck, netdev, gospo, Jeff Kirsher
In-Reply-To: <1316148173-11209-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Alexander Duyck <alexander.h.duyck@intel.com>
The allocation and freeing of the IRQ affinity hint needs some updates
since there are a number of spots where we run into possible issues with
the hint not being correctly updated.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 76 ++++++++++++-------------
1 files changed, 36 insertions(+), 40 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index 3ce0277..73a669d 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -1565,20 +1565,6 @@ static void ixgbe_configure_msix(struct ixgbe_adapter *adapter)
q_vector->eitr = adapter->rx_eitr_param;
ixgbe_write_eitr(q_vector);
- /* If ATR is enabled, set interrupt affinity */
- if (adapter->flags & IXGBE_FLAG_FDIR_HASH_CAPABLE) {
- /*
- * Allocate the affinity_hint cpumask, assign the mask
- * for this vector, and set our affinity_hint for
- * this irq.
- */
- if (!alloc_cpumask_var(&q_vector->affinity_mask,
- GFP_KERNEL))
- return;
- cpumask_set_cpu(v_idx, q_vector->affinity_mask);
- irq_set_affinity_hint(adapter->msix_entries[v_idx].vector,
- q_vector->affinity_mask);
- }
}
switch (adapter->hw.mac.type) {
@@ -2093,18 +2079,17 @@ out:
static int ixgbe_request_msix_irqs(struct ixgbe_adapter *adapter)
{
struct net_device *netdev = adapter->netdev;
- int i, vector, q_vectors, err;
+ int q_vectors = adapter->num_msix_vectors - NON_Q_VECTORS;
+ int vector, err;
int ri = 0, ti = 0;
- /* Decrement for Other and TCP Timer vectors */
- q_vectors = adapter->num_msix_vectors - NON_Q_VECTORS;
-
err = ixgbe_map_rings_to_vectors(adapter);
if (err)
return err;
for (vector = 0; vector < q_vectors; vector++) {
struct ixgbe_q_vector *q_vector = adapter->q_vector[vector];
+ struct msix_entry *entry = &adapter->msix_entries[vector];
if (q_vector->tx.ring && q_vector->rx.ring) {
snprintf(q_vector->name, sizeof(q_vector->name) - 1,
@@ -2120,14 +2105,19 @@ static int ixgbe_request_msix_irqs(struct ixgbe_adapter *adapter)
/* skip this unused q_vector */
continue;
}
- err = request_irq(adapter->msix_entries[vector].vector,
- &ixgbe_msix_clean_rings, 0, q_vector->name,
- q_vector);
+ err = request_irq(entry->vector, &ixgbe_msix_clean_rings, 0,
+ q_vector->name, q_vector);
if (err) {
e_err(probe, "request_irq failed for MSIX interrupt "
"Error: %d\n", err);
goto free_queue_irqs;
}
+ /* If Flow Director is enabled, set interrupt affinity */
+ if (adapter->flags & IXGBE_FLAG_FDIR_HASH_CAPABLE) {
+ /* assign the mask for this irq */
+ irq_set_affinity_hint(entry->vector,
+ q_vector->affinity_mask);
+ }
}
sprintf(adapter->lsc_int_name, "%s:lsc", netdev->name);
@@ -2141,9 +2131,13 @@ static int ixgbe_request_msix_irqs(struct ixgbe_adapter *adapter)
return 0;
free_queue_irqs:
- for (i = vector - 1; i >= 0; i--)
- free_irq(adapter->msix_entries[--vector].vector,
- adapter->q_vector[i]);
+ while (vector) {
+ vector--;
+ irq_set_affinity_hint(adapter->msix_entries[vector].vector,
+ NULL);
+ free_irq(adapter->msix_entries[vector].vector,
+ adapter->q_vector[vector]);
+ }
adapter->flags &= ~IXGBE_FLAG_MSIX_ENABLED;
pci_disable_msix(adapter->pdev);
kfree(adapter->msix_entries);
@@ -2333,14 +2327,19 @@ static void ixgbe_free_irq(struct ixgbe_adapter *adapter)
!adapter->q_vector[i]->tx.ring)
continue;
+ /* clear the affinity_mask in the IRQ descriptor */
+ irq_set_affinity_hint(adapter->msix_entries[i].vector,
+ NULL);
+
free_irq(adapter->msix_entries[i].vector,
adapter->q_vector[i]);
}
-
- ixgbe_reset_q_vectors(adapter);
} else {
free_irq(adapter->pdev->irq, adapter);
}
+
+ /* clear q_vector state information */
+ ixgbe_reset_q_vectors(adapter);
}
/**
@@ -3879,7 +3878,6 @@ void ixgbe_down(struct ixgbe_adapter *adapter)
struct ixgbe_hw *hw = &adapter->hw;
u32 rxctrl;
int i;
- int num_q_vectors = adapter->num_msix_vectors - NON_Q_VECTORS;
/* signal that we are down to the interrupt handler */
set_bit(__IXGBE_DOWN, &adapter->state);
@@ -3924,15 +3922,6 @@ void ixgbe_down(struct ixgbe_adapter *adapter)
adapter->vfinfo[i].clear_to_send = 0;
}
- /* Cleanup the affinity_hint CPU mask memory and callback */
- for (i = 0; i < num_q_vectors; i++) {
- struct ixgbe_q_vector *q_vector = adapter->q_vector[i];
- /* clear the affinity_mask in the IRQ descriptor */
- irq_set_affinity_hint(adapter->msix_entries[i]. vector, NULL);
- /* release the CPU mask memory */
- free_cpumask_var(q_vector->affinity_mask);
- }
-
/* disable transmits in the hardware now that interrupts are off */
for (i = 0; i < adapter->num_tx_queues; i++) {
u8 reg_idx = adapter->tx_ring[i]->reg_idx;
@@ -4677,6 +4666,11 @@ static int ixgbe_alloc_q_vectors(struct ixgbe_adapter *adapter)
q_vector->adapter = adapter;
q_vector->v_idx = v_idx;
+ /* Allocate the affinity_hint cpumask, configure the mask */
+ if (!alloc_cpumask_var(&q_vector->affinity_mask, GFP_KERNEL))
+ goto err_out;
+ cpumask_set_cpu(v_idx, q_vector->affinity_mask);
+
if (q_vector->tx.count && !q_vector->rx.count)
q_vector->eitr = adapter->tx_eitr_param;
else
@@ -4694,6 +4688,7 @@ err_out:
v_idx--;
q_vector = adapter->q_vector[v_idx];
netif_napi_del(&q_vector->napi);
+ free_cpumask_var(q_vector->affinity_mask);
kfree(q_vector);
adapter->q_vector[v_idx] = NULL;
}
@@ -4710,17 +4705,18 @@ err_out:
**/
static void ixgbe_free_q_vectors(struct ixgbe_adapter *adapter)
{
- int q_idx, num_q_vectors;
+ int v_idx, num_q_vectors;
if (adapter->flags & IXGBE_FLAG_MSIX_ENABLED)
num_q_vectors = adapter->num_msix_vectors - NON_Q_VECTORS;
else
num_q_vectors = 1;
- for (q_idx = 0; q_idx < num_q_vectors; q_idx++) {
- struct ixgbe_q_vector *q_vector = adapter->q_vector[q_idx];
- adapter->q_vector[q_idx] = NULL;
+ for (v_idx = 0; v_idx < num_q_vectors; v_idx++) {
+ struct ixgbe_q_vector *q_vector = adapter->q_vector[v_idx];
+ adapter->q_vector[v_idx] = NULL;
netif_napi_del(&q_vector->napi);
+ free_cpumask_var(q_vector->affinity_mask);
kfree(q_vector);
}
}
--
1.7.6
^ permalink raw reply related
* [net-next 04/11] ixgbe: Use ring->dev instead of adapter->pdev->dev when updating DCA
From: Jeff Kirsher @ 2011-09-16 4:42 UTC (permalink / raw)
To: davem; +Cc: Alexander Duyck, netdev, gospo, Jeff Kirsher
In-Reply-To: <1316148173-11209-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Alexander Duyck <alexander.h.duyck@intel.com>
This change switches us over to using the ring->dev pointer instead of
having to use the adapter->pdev->dev reference. The advantage to this is
that it is a much shorter route to get the to final needed value.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 8 ++++----
1 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index 73a669d..0564c65 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -924,12 +924,12 @@ static void ixgbe_update_rx_dca(struct ixgbe_adapter *adapter,
switch (hw->mac.type) {
case ixgbe_mac_82598EB:
rxctrl &= ~IXGBE_DCA_RXCTRL_CPUID_MASK;
- rxctrl |= dca3_get_tag(&adapter->pdev->dev, cpu);
+ rxctrl |= dca3_get_tag(rx_ring->dev, cpu);
break;
case ixgbe_mac_82599EB:
case ixgbe_mac_X540:
rxctrl &= ~IXGBE_DCA_RXCTRL_CPUID_MASK_82599;
- rxctrl |= (dca3_get_tag(&adapter->pdev->dev, cpu) <<
+ rxctrl |= (dca3_get_tag(rx_ring->dev, cpu) <<
IXGBE_DCA_RXCTRL_CPUID_SHIFT_82599);
break;
default:
@@ -953,7 +953,7 @@ static void ixgbe_update_tx_dca(struct ixgbe_adapter *adapter,
case ixgbe_mac_82598EB:
txctrl = IXGBE_READ_REG(hw, IXGBE_DCA_TXCTRL(reg_idx));
txctrl &= ~IXGBE_DCA_TXCTRL_CPUID_MASK;
- txctrl |= dca3_get_tag(&adapter->pdev->dev, cpu);
+ txctrl |= dca3_get_tag(tx_ring->dev, cpu);
txctrl |= IXGBE_DCA_TXCTRL_DESC_DCA_EN;
IXGBE_WRITE_REG(hw, IXGBE_DCA_TXCTRL(reg_idx), txctrl);
break;
@@ -961,7 +961,7 @@ static void ixgbe_update_tx_dca(struct ixgbe_adapter *adapter,
case ixgbe_mac_X540:
txctrl = IXGBE_READ_REG(hw, IXGBE_DCA_TXCTRL_82599(reg_idx));
txctrl &= ~IXGBE_DCA_TXCTRL_CPUID_MASK_82599;
- txctrl |= (dca3_get_tag(&adapter->pdev->dev, cpu) <<
+ txctrl |= (dca3_get_tag(tx_ring->dev, cpu) <<
IXGBE_DCA_TXCTRL_CPUID_SHIFT_82599);
txctrl |= IXGBE_DCA_TXCTRL_DESC_DCA_EN;
IXGBE_WRITE_REG(hw, IXGBE_DCA_TXCTRL_82599(reg_idx), txctrl);
--
1.7.6
^ permalink raw reply related
* [net-next 06/11] ixgbe: Drop unnecessary adapter->hw dereference in loopback test setup
From: Jeff Kirsher @ 2011-09-16 4:42 UTC (permalink / raw)
To: davem; +Cc: Alexander Duyck, netdev, gospo, Jeff Kirsher
In-Reply-To: <1316148173-11209-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Alexander Duyck <alexander.h.duyck@intel.com>
This patch drops a set of unnecessary dereferences to the hardware structure
since we already have a local copy of the hardware pointer.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c | 18 +++++++++---------
1 files changed, 9 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
index 9c12b35..11e1d5c 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
@@ -1570,26 +1570,26 @@ static int ixgbe_setup_loopback_test(struct ixgbe_adapter *adapter)
/* X540 needs to set the MACC.FLU bit to force link up */
if (adapter->hw.mac.type == ixgbe_mac_X540) {
- reg_data = IXGBE_READ_REG(&adapter->hw, IXGBE_MACC);
+ reg_data = IXGBE_READ_REG(hw, IXGBE_MACC);
reg_data |= IXGBE_MACC_FLU;
- IXGBE_WRITE_REG(&adapter->hw, IXGBE_MACC, reg_data);
+ IXGBE_WRITE_REG(hw, IXGBE_MACC, reg_data);
}
/* right now we only support MAC loopback in the driver */
- reg_data = IXGBE_READ_REG(&adapter->hw, IXGBE_HLREG0);
+ reg_data = IXGBE_READ_REG(hw, IXGBE_HLREG0);
/* Setup MAC loopback */
reg_data |= IXGBE_HLREG0_LPBK;
- IXGBE_WRITE_REG(&adapter->hw, IXGBE_HLREG0, reg_data);
+ IXGBE_WRITE_REG(hw, IXGBE_HLREG0, reg_data);
- reg_data = IXGBE_READ_REG(&adapter->hw, IXGBE_FCTRL);
+ reg_data = IXGBE_READ_REG(hw, IXGBE_FCTRL);
reg_data |= IXGBE_FCTRL_BAM | IXGBE_FCTRL_SBP | IXGBE_FCTRL_MPE;
- IXGBE_WRITE_REG(&adapter->hw, IXGBE_FCTRL, reg_data);
+ IXGBE_WRITE_REG(hw, IXGBE_FCTRL, reg_data);
- reg_data = IXGBE_READ_REG(&adapter->hw, IXGBE_AUTOC);
+ reg_data = IXGBE_READ_REG(hw, IXGBE_AUTOC);
reg_data &= ~IXGBE_AUTOC_LMS_MASK;
reg_data |= IXGBE_AUTOC_LMS_10G_LINK_NO_AN | IXGBE_AUTOC_FLU;
- IXGBE_WRITE_REG(&adapter->hw, IXGBE_AUTOC, reg_data);
- IXGBE_WRITE_FLUSH(&adapter->hw);
+ IXGBE_WRITE_REG(hw, IXGBE_AUTOC, reg_data);
+ IXGBE_WRITE_FLUSH(hw);
usleep_range(10000, 20000);
/* Disable Atlas Tx lanes; re-enabled in reset path */
--
1.7.6
^ permalink raw reply related
* [net-next 07/11] ixgbe: combine PCI_VDEVICE and board declaration to same line
From: Jeff Kirsher @ 2011-09-16 4:42 UTC (permalink / raw)
To: davem; +Cc: Alexander Duyck, netdev, gospo, Jeff Kirsher
In-Reply-To: <1316148173-11209-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Alexander Duyck <alexander.h.duyck@intel.com>
This patch is a minor whitespace cleanup to compress the device ID
declaration and board type declaration onto the same line. It seems to
make sense since all of the combinations of the two are less than 80
characters and it makes the overall layout a bit more readable.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 79 ++++++++-----------------
1 files changed, 26 insertions(+), 53 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index dd1b57b..fee5630 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -79,59 +79,32 @@ static const struct ixgbe_info *ixgbe_info_tbl[] = {
* Class, Class Mask, private data (not used) }
*/
static DEFINE_PCI_DEVICE_TABLE(ixgbe_pci_tbl) = {
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598),
- board_82598 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598AF_DUAL_PORT),
- board_82598 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598AF_SINGLE_PORT),
- board_82598 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598AT),
- board_82598 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598AT2),
- board_82598 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598EB_CX4),
- board_82598 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598_CX4_DUAL_PORT),
- board_82598 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598_DA_DUAL_PORT),
- board_82598 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598_SR_DUAL_PORT_EM),
- board_82598 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598EB_XF_LR),
- board_82598 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598EB_SFP_LOM),
- board_82598 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598_BX),
- board_82598 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_KX4),
- board_82599 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_XAUI_LOM),
- board_82599 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_KR),
- board_82599 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_SFP),
- board_82599 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_SFP_EM),
- board_82599 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_KX4_MEZZ),
- board_82599 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_CX4),
- board_82599 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_BACKPLANE_FCOE),
- board_82599 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_SFP_FCOE),
- board_82599 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_T3_LOM),
- board_82599 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_COMBO_BACKPLANE),
- board_82599 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_X540T),
- board_X540 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_SFP_SF2),
- board_82599 },
- {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_LS),
- board_82599 },
-
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598), board_82598 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598AF_DUAL_PORT), board_82598 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598AF_SINGLE_PORT), board_82598 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598AT), board_82598 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598AT2), board_82598 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598EB_CX4), board_82598 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598_CX4_DUAL_PORT), board_82598 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598_DA_DUAL_PORT), board_82598 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598_SR_DUAL_PORT_EM), board_82598 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598EB_XF_LR), board_82598 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598EB_SFP_LOM), board_82598 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82598_BX), board_82598 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_KX4), board_82599 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_XAUI_LOM), board_82599 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_KR), board_82599 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_SFP), board_82599 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_SFP_EM), board_82599 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_KX4_MEZZ), board_82599 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_CX4), board_82599 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_BACKPLANE_FCOE), board_82599 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_SFP_FCOE), board_82599 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_T3_LOM), board_82599 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_COMBO_BACKPLANE), board_82599 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_X540T), board_X540 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_SFP_SF2), board_82599 },
+ {PCI_VDEVICE(INTEL, IXGBE_DEV_ID_82599_LS), board_82599 },
/* required last entry */
{0, }
};
--
1.7.6
^ permalink raw reply related
* [net-next 08/11] ixgbe: Update TXDCTL configuration to correctly handle WTHRESH
From: Jeff Kirsher @ 2011-09-16 4:42 UTC (permalink / raw)
To: davem; +Cc: Alexander Duyck, netdev, gospo, Jeff Kirsher
In-Reply-To: <1316148173-11209-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Alexander Duyck <alexander.h.duyck@intel.com>
This change updated the TXDCTL configuration. The main goal is to be much
more explicit about the configuration and avoid a possible fake TX hang
when the interrupt throttle rate is set to 0.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 35 +++++++++++++------------
1 files changed, 18 insertions(+), 17 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index fee5630..6378d7f 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -2359,13 +2359,11 @@ void ixgbe_configure_tx_ring(struct ixgbe_adapter *adapter,
struct ixgbe_hw *hw = &adapter->hw;
u64 tdba = ring->dma;
int wait_loop = 10;
- u32 txdctl;
+ u32 txdctl = IXGBE_TXDCTL_ENABLE;
u8 reg_idx = ring->reg_idx;
/* disable queue to avoid issues while updating state */
- txdctl = IXGBE_READ_REG(hw, IXGBE_TXDCTL(reg_idx));
- IXGBE_WRITE_REG(hw, IXGBE_TXDCTL(reg_idx),
- txdctl & ~IXGBE_TXDCTL_ENABLE);
+ IXGBE_WRITE_REG(hw, IXGBE_TXDCTL(reg_idx), 0);
IXGBE_WRITE_FLUSH(hw);
IXGBE_WRITE_REG(hw, IXGBE_TDBAL(reg_idx),
@@ -2377,18 +2375,22 @@ void ixgbe_configure_tx_ring(struct ixgbe_adapter *adapter,
IXGBE_WRITE_REG(hw, IXGBE_TDT(reg_idx), 0);
ring->tail = hw->hw_addr + IXGBE_TDT(reg_idx);
- /* configure fetching thresholds */
- if (adapter->rx_itr_setting == 0) {
- /* cannot set wthresh when itr==0 */
- txdctl &= ~0x007F0000;
- } else {
- /* enable WTHRESH=8 descriptors, to encourage burst writeback */
- txdctl |= (8 << 16);
- }
- if (adapter->flags & IXGBE_FLAG_DCB_ENABLED) {
- /* PThresh workaround for Tx hang with DFP enabled. */
- txdctl |= 32;
- }
+ /*
+ * set WTHRESH to encourage burst writeback, it should not be set
+ * higher than 1 when ITR is 0 as it could cause false TX hangs
+ *
+ * In order to avoid issues WTHRESH + PTHRESH should always be equal
+ * to or less than the number of on chip descriptors, which is
+ * currently 40.
+ */
+ if (!adapter->tx_itr_setting || !adapter->rx_itr_setting)
+ txdctl |= (1 << 16); /* WTHRESH = 1 */
+ else
+ txdctl |= (8 << 16); /* WTHRESH = 8 */
+
+ /* PTHRESH=32 is needed to avoid a Tx hang with DFP enabled. */
+ txdctl |= (1 << 8) | /* HTHRESH = 1 */
+ 32; /* PTHRESH = 32 */
/* reinitialize flowdirector state */
if ((adapter->flags & IXGBE_FLAG_FDIR_HASH_CAPABLE) &&
@@ -2403,7 +2405,6 @@ void ixgbe_configure_tx_ring(struct ixgbe_adapter *adapter,
clear_bit(__IXGBE_HANG_CHECK_ARMED, &ring->state);
/* enable queue */
- txdctl |= IXGBE_TXDCTL_ENABLE;
IXGBE_WRITE_REG(hw, IXGBE_TXDCTL(reg_idx), txdctl);
/* TXDCTL.EN will return 0 on 82598 if link is down, so skip it */
--
1.7.6
^ permalink raw reply related
* [net-next 09/11] ixgbe: cleanup reset paths
From: Jeff Kirsher @ 2011-09-16 4:42 UTC (permalink / raw)
To: davem; +Cc: Alexander Duyck, netdev, gospo, Jeff Kirsher
In-Reply-To: <1316148173-11209-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Alexander Duyck <alexander.h.duyck@intel.com>
The reset paths are overly complicated and are either missing steps or
contain extra unnecessary steps such as reading MAC address twice. This
change is meant to help clean up the reset paths an get things functioning
correctly.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe_82598.c | 13 ++---
drivers/net/ethernet/intel/ixgbe/ixgbe_82599.c | 40 ++++++++------
drivers/net/ethernet/intel/ixgbe/ixgbe_type.h | 1 +
drivers/net/ethernet/intel/ixgbe/ixgbe_x540.c | 72 +++++-------------------
4 files changed, 43 insertions(+), 83 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_82598.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_82598.c
index 0d4e382..22504f2 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_82598.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_82598.c
@@ -820,8 +820,8 @@ mac_reset_top:
* Issue global reset to the MAC. This needs to be a SW reset.
* If link reset is used, it might reset the MAC when mng is using it
*/
- ctrl = IXGBE_READ_REG(hw, IXGBE_CTRL);
- IXGBE_WRITE_REG(hw, IXGBE_CTRL, (ctrl | IXGBE_CTRL_RST));
+ ctrl = IXGBE_READ_REG(hw, IXGBE_CTRL) | IXGBE_CTRL_RST;
+ IXGBE_WRITE_REG(hw, IXGBE_CTRL, ctrl);
IXGBE_WRITE_FLUSH(hw);
/* Poll for reset bit to self-clear indicating reset is complete */
@@ -836,21 +836,18 @@ mac_reset_top:
hw_dbg(hw, "Reset polling failed to complete.\n");
}
+ msleep(50);
+
/*
* Double resets are required for recovery from certain error
* conditions. Between resets, it is necessary to stall to allow time
- * for any pending HW events to complete. We use 1usec since that is
- * what is needed for ixgbe_disable_pcie_master(). The second reset
- * then clears out any effects of those events.
+ * for any pending HW events to complete.
*/
if (hw->mac.flags & IXGBE_FLAGS_DOUBLE_RESET_REQUIRED) {
hw->mac.flags &= ~IXGBE_FLAGS_DOUBLE_RESET_REQUIRED;
- udelay(1);
goto mac_reset_top;
}
- msleep(50);
-
gheccr = IXGBE_READ_REG(hw, IXGBE_GHECCR);
gheccr &= ~((1 << 21) | (1 << 18) | (1 << 9) | (1 << 6));
IXGBE_WRITE_REG(hw, IXGBE_GHECCR, gheccr);
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_82599.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_82599.c
index f193fc2..a5ff435 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_82599.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_82599.c
@@ -904,11 +904,10 @@ static s32 ixgbe_setup_copper_link_82599(struct ixgbe_hw *hw,
**/
static s32 ixgbe_reset_hw_82599(struct ixgbe_hw *hw)
{
- s32 status = 0;
- u32 ctrl;
- u32 i;
- u32 autoc;
- u32 autoc2;
+ ixgbe_link_speed link_speed;
+ s32 status;
+ u32 ctrl, i, autoc, autoc2;
+ bool link_up = false;
/* Call adapter stop to disable tx/rx and clear interrupts */
hw->mac.ops.stop_adapter(hw);
@@ -942,40 +941,47 @@ static s32 ixgbe_reset_hw_82599(struct ixgbe_hw *hw)
mac_reset_top:
/*
- * Issue global reset to the MAC. This needs to be a SW reset.
- * If link reset is used, it might reset the MAC when mng is using it
+ * Issue global reset to the MAC. Needs to be SW reset if link is up.
+ * If link reset is used when link is up, it might reset the PHY when
+ * mng is using it. If link is down or the flag to force full link
+ * reset is set, then perform link reset.
*/
- ctrl = IXGBE_READ_REG(hw, IXGBE_CTRL);
- IXGBE_WRITE_REG(hw, IXGBE_CTRL, (ctrl | IXGBE_CTRL_RST));
+ ctrl = IXGBE_CTRL_LNK_RST;
+ if (!hw->force_full_reset) {
+ hw->mac.ops.check_link(hw, &link_speed, &link_up, false);
+ if (link_up)
+ ctrl = IXGBE_CTRL_RST;
+ }
+
+ ctrl |= IXGBE_READ_REG(hw, IXGBE_CTRL);
+ IXGBE_WRITE_REG(hw, IXGBE_CTRL, ctrl);
IXGBE_WRITE_FLUSH(hw);
/* Poll for reset bit to self-clear indicating reset is complete */
for (i = 0; i < 10; i++) {
udelay(1);
ctrl = IXGBE_READ_REG(hw, IXGBE_CTRL);
- if (!(ctrl & IXGBE_CTRL_RST))
+ if (!(ctrl & IXGBE_CTRL_RST_MASK))
break;
}
- if (ctrl & IXGBE_CTRL_RST) {
+
+ if (ctrl & IXGBE_CTRL_RST_MASK) {
status = IXGBE_ERR_RESET_FAILED;
hw_dbg(hw, "Reset polling failed to complete.\n");
}
+ msleep(50);
+
/*
* Double resets are required for recovery from certain error
* conditions. Between resets, it is necessary to stall to allow time
- * for any pending HW events to complete. We use 1usec since that is
- * what is needed for ixgbe_disable_pcie_master(). The second reset
- * then clears out any effects of those events.
+ * for any pending HW events to complete.
*/
if (hw->mac.flags & IXGBE_FLAGS_DOUBLE_RESET_REQUIRED) {
hw->mac.flags &= ~IXGBE_FLAGS_DOUBLE_RESET_REQUIRED;
- udelay(1);
goto mac_reset_top;
}
- msleep(50);
-
/*
* Store the original AUTOC/AUTOC2 values if they have not been
* stored off yet. Otherwise restore the stored original
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h b/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h
index 9f618ee..a9f8839 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h
@@ -982,6 +982,7 @@
#define IXGBE_CTRL_GIO_DIS 0x00000004 /* Global IO Master Disable bit */
#define IXGBE_CTRL_LNK_RST 0x00000008 /* Link Reset. Resets everything. */
#define IXGBE_CTRL_RST 0x04000000 /* Reset (SW) */
+#define IXGBE_CTRL_RST_MASK (IXGBE_CTRL_LNK_RST | IXGBE_CTRL_RST)
/* FACTPS */
#define IXGBE_FACTPS_LFS 0x40000000 /* LAN Function Select */
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_x540.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_x540.c
index 2696c78..bbfe8c4 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_x540.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_x540.c
@@ -94,13 +94,8 @@ static s32 ixgbe_setup_mac_link_X540(struct ixgbe_hw *hw,
static s32 ixgbe_reset_hw_X540(struct ixgbe_hw *hw)
{
ixgbe_link_speed link_speed;
- s32 status = 0;
- u32 ctrl;
- u32 ctrl_ext;
- u32 reset_bit;
- u32 i;
- u32 autoc;
- u32 autoc2;
+ s32 status;
+ u32 ctrl, i;
bool link_up = false;
/* Call adapter stop to disable tx/rx and clear interrupts */
@@ -119,53 +114,42 @@ mac_reset_top:
* mng is using it. If link is down or the flag to force full link
* reset is set, then perform link reset.
*/
- if (hw->force_full_reset) {
- reset_bit = IXGBE_CTRL_LNK_RST;
- } else {
+ ctrl = IXGBE_CTRL_LNK_RST;
+ if (!hw->force_full_reset) {
hw->mac.ops.check_link(hw, &link_speed, &link_up, false);
- if (!link_up)
- reset_bit = IXGBE_CTRL_LNK_RST;
- else
- reset_bit = IXGBE_CTRL_RST;
+ if (link_up)
+ ctrl = IXGBE_CTRL_RST;
}
- ctrl = IXGBE_READ_REG(hw, IXGBE_CTRL);
- IXGBE_WRITE_REG(hw, IXGBE_CTRL, (ctrl | reset_bit));
+ ctrl |= IXGBE_READ_REG(hw, IXGBE_CTRL);
+ IXGBE_WRITE_REG(hw, IXGBE_CTRL, ctrl);
IXGBE_WRITE_FLUSH(hw);
/* Poll for reset bit to self-clear indicating reset is complete */
for (i = 0; i < 10; i++) {
udelay(1);
ctrl = IXGBE_READ_REG(hw, IXGBE_CTRL);
- if (!(ctrl & reset_bit))
+ if (!(ctrl & IXGBE_CTRL_RST_MASK))
break;
}
- if (ctrl & reset_bit) {
+
+ if (ctrl & IXGBE_CTRL_RST_MASK) {
status = IXGBE_ERR_RESET_FAILED;
hw_dbg(hw, "Reset polling failed to complete.\n");
}
+ msleep(50);
+
/*
* Double resets are required for recovery from certain error
* conditions. Between resets, it is necessary to stall to allow time
- * for any pending HW events to complete. We use 1usec since that is
- * what is needed for ixgbe_disable_pcie_master(). The second reset
- * then clears out any effects of those events.
+ * for any pending HW events to complete.
*/
if (hw->mac.flags & IXGBE_FLAGS_DOUBLE_RESET_REQUIRED) {
hw->mac.flags &= ~IXGBE_FLAGS_DOUBLE_RESET_REQUIRED;
- udelay(1);
goto mac_reset_top;
}
- /* Clear PF Reset Done bit so PF/VF Mail Ops can work */
- ctrl_ext = IXGBE_READ_REG(hw, IXGBE_CTRL_EXT);
- ctrl_ext |= IXGBE_CTRL_EXT_PFRSTD;
- IXGBE_WRITE_REG(hw, IXGBE_CTRL_EXT, ctrl_ext);
- IXGBE_WRITE_FLUSH(hw);
-
- msleep(50);
-
/* Set the Rx packet buffer size. */
IXGBE_WRITE_REG(hw, IXGBE_RXPBSIZE(0), 384 << IXGBE_RXPBSIZE_SHIFT);
@@ -173,31 +157,6 @@ mac_reset_top:
hw->mac.ops.get_mac_addr(hw, hw->mac.perm_addr);
/*
- * Store the original AUTOC/AUTOC2 values if they have not been
- * stored off yet. Otherwise restore the stored original
- * values since the reset operation sets back to defaults.
- */
- autoc = IXGBE_READ_REG(hw, IXGBE_AUTOC);
- autoc2 = IXGBE_READ_REG(hw, IXGBE_AUTOC2);
- if (hw->mac.orig_link_settings_stored == false) {
- hw->mac.orig_autoc = autoc;
- hw->mac.orig_autoc2 = autoc2;
- hw->mac.orig_link_settings_stored = true;
- } else {
- if (autoc != hw->mac.orig_autoc)
- IXGBE_WRITE_REG(hw, IXGBE_AUTOC, (hw->mac.orig_autoc |
- IXGBE_AUTOC_AN_RESTART));
-
- if ((autoc2 & IXGBE_AUTOC2_UPPER_MASK) !=
- (hw->mac.orig_autoc2 & IXGBE_AUTOC2_UPPER_MASK)) {
- autoc2 &= ~IXGBE_AUTOC2_UPPER_MASK;
- autoc2 |= (hw->mac.orig_autoc2 &
- IXGBE_AUTOC2_UPPER_MASK);
- IXGBE_WRITE_REG(hw, IXGBE_AUTOC2, autoc2);
- }
- }
-
- /*
* Store MAC address from RAR0, clear receive address registers, and
* clear the multicast table. Also reset num_rar_entries to 128,
* since we modify this value when programming the SAN MAC address.
@@ -205,9 +164,6 @@ mac_reset_top:
hw->mac.num_rar_entries = IXGBE_X540_MAX_TX_QUEUES;
hw->mac.ops.init_rx_addrs(hw);
- /* Store the permanent mac address */
- hw->mac.ops.get_mac_addr(hw, hw->mac.perm_addr);
-
/* Store the permanent SAN mac address */
hw->mac.ops.get_san_mac_addr(hw, hw->mac.san_addr);
--
1.7.6
^ 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