Skip to contentSkip to conversation

TUN/TAP driver update and fixes for 2.6.BK

BK Max Krasnyansky 3 participants 9 messages Jan 12, 2005 – Jan 13, 2005

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Dave, Andrew,

Could one of you please pull TUN/TAP driver updates from my tree

        bk://maxk.bkbits.net/tun-2.6

This will update the following files:

 drivers/net/Kconfig    |    1
 drivers/net/tun.c      |  145 ++++++++++++++++++++++++++++++++++++++++++-------
 include/linux/if_tun.h |    2
 3 files changed, 128 insertions(+), 20 deletions(-)

through these ChangeSets:

<[email protected]> (05/01/12 1.2379)
[TUN] Add a missing dependency on enabling the crc32 libraries

Patch by Steve French <[email protected]>

<[email protected]> (04/12/23 1.1938.458.6)
Use random_ether_addr() to generate TAP MAC address.

<[email protected]> (04/12/23 1.1938.458.5)
TUN/TAP driver packet queuing fixes and improvements

Patch from Harald Roelle <[email protected]>

Fixes for the following issues
- "Kicking" packet behavior in case of kernel packet scheduler (! TUN_ONE_QUEUE):
When the netif queue is stopped because of an overrun of the driver's queue, no new packets are delivered to the driver until a new packet arrives, not even when in the meantime there's again space in the driver's queue (gained by a reading user process). In short, whenever netif queue was stopped, one needs an additional packet to "kick" packet delivery to the driver.
The reason for this is, that the netif queue is started but not woken up, i.e. the rest of the kernel is not signaled to resume packet delivery.

- Adjustment of tx queue length by ifconfig has only effect when TUN_ONE_QUEUE is set. Otherwise always constant TUN_READQ_SIZE queue length is used.

- TX queue default length setting is not consistent:
o TAP dev + TUN_ONE_QUEUE: 1000 (by ether_setup())
o all other cases: 10

- Default tx queue length is too short in many cases.
IMHO it should be at least twice as long as the number of fragments needed for a maximum sized IP packet at a "typical" MTU size.
This would ensure that at least one complete IP packet can be delivered to the attached user space process, even if the packet's fragments are "misaligned" in the buffer and the user process is not scheduled in time to read the queue.

Additional modifications:

- To signal that stopping of the queue has occurred, the tx fifo overrun counter is increased.

- Implemented ethtool api. Primarily added to have a standard method requesting the driver version.

Thanks
Max

#1 lore raw
Quoted 2 lines · Max Krasnyansky

Could one of you please pull TUN/TAP driver updates from my tree

That would be Dave.

Did you see this fly past on netdev?

From: Tommy Christensen <[email protected]>

But as stated in bonding.txt, the ARP monitor requires the underlying driver to update dev->trans_start and dev->last_rx.

The patch below adds the required functionality to the TUN/TAP driver. Please test if this helps in your case.

---

 25-akpm/drivers/net/tun.c |    2 ++
 1 files changed, 2 insertions(+)
drivers/net/tun.c renamed from drivers/net/tun.c~tun-tan-arp-monitor-support +2 -0
diff -puN drivers/net/tun.c~tun-tan-arp-monitor-support drivers/net/tun.c
@@ -92,6 +92,7 @@ static int tun_net_xmit(struct sk_buff *
 			goto drop;
 	}
 	skb_queue_tail(&tun->readq, skb);
+	dev->trans_start = jiffies;
 
 	/* Notify and wake up reader process */
 	if (tun->flags & TUN_FASYNC)
@@ -240,6 +241,7 @@ static __inline__ ssize_t tun_get_user(s
 		skb->ip_summed = CHECKSUM_UNNECESSARY;
  
 	netif_rx_ni(skb);
+	tun->dev->last_rx = jiffies;
    
 	tun->stats.rx_packets++;
 	tun->stats.rx_bytes += len;

_

#2 lore raw
Quoted 6 lines · Andrew Morton
Quoted 1 line · Max Krasnyansky

Could one of you please pull TUN/TAP driver updates from my tree

That would be Dave.

or me, being drivers/net :) I pulled it.

Comments coming in separate email.

Jeff

#3 lore raw
In reply to Andrew Morton
Quoted 7 lines · Andrew Morton
Quoted 1 line · Max Krasnyansky

Could one of you please pull TUN/TAP driver updates from my tree

That would be Dave.

Did you see this fly past on netdev?

Nope. Somehow I'm not getting any mail on netdev. I guess I need to check my subscription :).

I'll apply it an push to Dave.

Thanks
Max

#4 lore raw
Quoted 6 lines · Max Krasnyansky
Quoted 4 lines · Andrew Morton
Quoted 1 line · Max Krasnyansky

Could one of you please pull TUN/TAP driver updates from my tree

Quoted 7 lines
Quoted 3 lines

That would be Dave.

Did you see this fly past on netdev?

Nope. Somehow I'm not getting any mail on netdev. I guess I need to check my subscription :).

I'll apply it an push to Dave.

Well. As far as I know we don't know if it works yet.

#5 lore raw
Quoted 1 line · Andrew Morton

Well. As far as I know we don't know if it works yet.

It looks totally correct to me. Even if it does not fix ARP monitor problem :).

Max

#6 lore raw
In reply to Max Krasnyansky
Quoted 5 lines · Max Krasnyansky

Dave, Andrew,

Could one of you please pull TUN/TAP driver updates from my tree

        bk://maxk.bkbits.net/tun-2.6
Quoted 7 lines

This will update the following files:

 drivers/net/Kconfig    |    1
 drivers/net/tun.c      |  145 
++++++++++++++++++++++++++++++++++++++++++-------
 include/linux/if_tun.h |    2
 3 files changed, 128 insertions(+), 20 deletions(-)

Non-technical comments:

1) Please send drivers/net patches to me and [email protected]

2) Consider using the bk-make-sum script (in Documentation/BK-usage/) to generate your summary. This will add a "bk pull " prefix to your BK url particularly, making it even easier to cut-n-paste.

3) Please include a patch in your submission so that list readers may review your changes, not just the BK users.

Technical comments:

1) Accepted, I pulled it into my netdev-2.6 queue

2) in your implementation of tun_get_drvinfo(), it may be nice to include the tun/tap interface number in info->bus_info, to differentiate between multiple tun interfaces or multiple tap interfaces.

3) You might consider moving tun_set_msglevel() completely inside TUN_DEBUG ifdef.

4) use of MODULE_VERSION() is recommended

#7 lore raw
Quoted 3 lines · Jeff Garzik

Non-technical comments:

1) Please send drivers/net patches to me and [email protected]

Ok

Quoted 1 line

2) Consider using the bk-make-sum script (in Documentation/BK-usage/) to generate your summary. This will add a "bk pull " prefix to your BK url particularly, making it even easier to cut-n-paste.

I do use bk-make-sum. A bit hacked version though which does not add 'bk pull' prefix. I'll put it back in if it's useful for folks.

Quoted 1 line

3) Please include a patch in your submission so that list readers may review your changes, not just the BK users.

Anybody can go to bkbits.net and review them. I'd rather not send patches along with BK stuff, unless that's a new rule or something :).

Quoted 7 lines

Technical comments:

2) in your implementation of tun_get_drvinfo(), it may be nice to include the tun/tap interface number in info->bus_info, to differentiate between multiple tun interfaces or multiple tap interfaces.

3) You might consider moving tun_set_msglevel() completely inside TUN_DEBUG ifdef.

4) use of MODULE_VERSION() is recommended

Good points.

Thanks
Max

#8 lore raw
Quoted 10 lines · Max Krasnyansky
Quoted 3 lines · Jeff Garzik

Non-technical comments:

1) Please send drivers/net patches to me and [email protected]

Ok

Quoted 1 line

2) Consider using the bk-make-sum script (in Documentation/BK-usage/) to generate your summary. This will add a "bk pull " prefix to your BK url particularly, making it even easier to cut-n-paste.

Quoted 7 lines

I do use bk-make-sum. A bit hacked version though which does not add 'bk pull' prefix. I'll put it back in if it's useful for folks.

Quoted 1 line

3) Please include a patch in your submission so that list readers may review your changes, not just the BK users.

Anybody can go to bkbits.net and review them. I'd rather not send patches along with BK stuff, unless that's a new rule or something :).

It's not a new rule, it's an old rule :) Changes need exposure and review _before_ they get merged and hit the kernel. This is part of how Linux has always worked.

It's trivial to generate such a patch and include it in your email, using the gcapatch script in Documentation/BK-usage/

Jeff