Page MenuHomeFreeBSD

jacob.e.keller_intel.com (Jacob Keller)
User

Projects

User Details

User Since
Aug 24 2017, 3:21 PM (68 w, 1 d)

Recent Activity

Yesterday

jacob.e.keller_intel.com added a comment to D18545: intel: correct initialization of tx_cidx_processed.

Stephen Hurd, this is what I believe to be the correct fix for this issue. I'm not entirely certain how to reproduce the original setup that triggered the bug, so some help with that would be appreciated.

Fri, Dec 14, 12:04 AM
jacob.e.keller_intel.com created D18545: intel: correct initialization of tx_cidx_processed.
Fri, Dec 14, 12:00 AM

Thu, Dec 6

jacob.e.keller_intel.com added inline comments to D18368: Fix first-packet completion.
Thu, Dec 6, 11:10 PM
jacob.e.keller_intel.com added a comment to D18368: Fix first-packet completion.

I know I'm late commenting on this, but I think the correct fix is to initialize tx_cidx_processed to txr->desc_count -1 when we initialize.

Thu, Dec 6, 11:10 PM

Nov 7 2018

jacob.e.keller_intel.com added reviewers for D17900: ixl/iavf: fix TSO offloads when TXCSUM is disabled: Intel Networking, shurd, erj.
Nov 7 2018, 11:56 PM
jacob.e.keller_intel.com created D17900: ixl/iavf: fix TSO offloads when TXCSUM is disabled.
Nov 7 2018, 11:54 PM
jacob.e.keller_intel.com added inline comments to D17881: Fix rxcsum issue introduced in r338838.
Nov 7 2018, 9:31 PM
jacob.e.keller_intel.com added a comment to D17881: Fix rxcsum issue introduced in r338838.

@shurd - Turns out you've been faster. We've been working one this bug too. Our plan was to do it this way:

			if ((setmask & (IFCAP_RXCSUM | IFCAP_RXCSUM_IPV6)) !=
			    (oldmask & (IFCAP_RXCSUM | IFCAP_RXCSUM_IPV6)))
				setmask |= IFCAP_RXCSUM | IFCAP_RXCSUM_IPV6;

But it's too late for me today to figure out which has more sense ;)

Nov 7 2018, 9:29 PM

Nov 5 2018

jacob.e.keller_intel.com added a comment to D17801: TSO inherently requires checksum offload. In the case where TSO is in use, force the checksum offload on as well for that packet..

If I fix the driver to enable IP checksums on TSOs even when CSUM_IP isn't set, *and* I enable IFLIB_TSO_INIT_IP, then this patch isn't necessary.

Nov 5 2018, 11:12 PM
jacob.e.keller_intel.com added a comment to D17801: TSO inherently requires checksum offload. In the case where TSO is in use, force the checksum offload on as well for that packet..

Actually, I think the reason we set NEED_CSUM_ZERO in some hardware is only because they need the IP sum zero'd during TSO, not because they need it for non-TSO packets....

Nov 5 2018, 11:05 PM

Nov 2 2018

jacob.e.keller_intel.com added a comment to D17801: TSO inherently requires checksum offload. In the case where TSO is in use, force the checksum offload on as well for that packet..

Thanks! This looks great.

Nov 2 2018, 9:17 PM
jacob.e.keller_intel.com added a comment to D17801: TSO inherently requires checksum offload. In the case where TSO is in use, force the checksum offload on as well for that packet..

So this is a step in the right direction, but isn't a complete fix for the issue. We also need to enable the ip_sum workaround in this case as well. Possibly we can just move that check to after the TX_OFFLOAD4 and TX_OFFLOAD6 checks?

Nov 2 2018, 7:43 PM
jacob.e.keller_intel.com added inline comments to D17801: TSO inherently requires checksum offload. In the case where TSO is in use, force the checksum offload on as well for that packet..
Nov 2 2018, 7:12 PM
jacob.e.keller_intel.com added a comment to D17801: TSO inherently requires checksum offload. In the case where TSO is in use, force the checksum offload on as well for that packet..

This doesn't seem like the right fix.

Nov 2 2018, 7:11 PM

May 11 2018

jacob.e.keller_intel.com updated the diff for D15343: iflib: mark irq allocation name parameter as constant.
  • taskqgroup: make tqg_name a constant
  • taskqgroup: mark name paramter of taskqgroup_attach as const
  • iflib: mark irq allocation name parameter as constant
May 11 2018, 6:50 PM
jacob.e.keller_intel.com added a comment to D15343: iflib: mark irq allocation name parameter as constant.

Looks like this involves a bit more cleanup to the taskqgroup code (since ultimately the name gets passed there). due to the scope I think I'll split it into two commits.

May 11 2018, 6:22 PM

May 8 2018

jacob.e.keller_intel.com added a comment to D15343: iflib: mark irq allocation name parameter as constant.

Hmmm... I think there were. I'll double check this later today and see what I missed. I have a separate system for sending the reviews because my test system doesn't have direct access, I probably just forgot a few lines..

May 8 2018, 5:28 PM

May 7 2018

jacob.e.keller_intel.com created D15343: iflib: mark irq allocation name parameter as constant.
May 7 2018, 11:15 PM

May 4 2018

jacob.e.keller_intel.com updated the summary of D15300: iflib: print message when iflib_tx_structures_setup fails.
May 4 2018, 11:33 PM
jacob.e.keller_intel.com updated the diff for D15300: iflib: print message when iflib_tx_structures_setup fails.

Remove the redundant message in iflib_device_register

May 4 2018, 11:32 PM
jacob.e.keller_intel.com abandoned D15304: iflib: print message when iflib_tx_structures_setup fails.

Oops didn't mean to upload a new revision..

May 4 2018, 11:31 PM
jacob.e.keller_intel.com created D15304: iflib: print message when iflib_tx_structures_setup fails.
May 4 2018, 11:30 PM
jacob.e.keller_intel.com added a comment to D15299: iflib: cleanup queues when iflib_device_register fail.

While we're cleaning this up, I'd either remove this print, or add a matching one for the tx setup, and remove the one in the caller. I realize that you can deduce which failed by how many prints you get, but that seems ... weird..

May 4 2018, 8:41 PM
jacob.e.keller_intel.com added a comment to D15300: iflib: print message when iflib_tx_structures_setup fails.

As mentioned on the other review, maybe remove the print in the caller, since it is now redundant..

May 4 2018, 8:40 PM
jacob.e.keller_intel.com created D15300: iflib: print message when iflib_tx_structures_setup fails.
May 4 2018, 6:53 PM
jacob.e.keller_intel.com created D15299: iflib: cleanup queues when iflib_device_register fail.
May 4 2018, 6:52 PM
jacob.e.keller_intel.com added a comment to D15285: iflib: fix invalid free during queue allocation failure.

Can we MFC this to stable/11? I originally found the bug on my FreeBSD 11.1 system.

May 4 2018, 6:30 PM

May 3 2018

jacob.e.keller_intel.com added a child revision for D15284: iflib: remove unused brscp pointer from iflib_queues_alloc: D15285: iflib: fix invalid free during queue allocation failure.
May 3 2018, 9:43 PM
jacob.e.keller_intel.com added a parent revision for D15285: iflib: fix invalid free during queue allocation failure: D15284: iflib: remove unused brscp pointer from iflib_queues_alloc.
May 3 2018, 9:43 PM
jacob.e.keller_intel.com added a comment to D15285: iflib: fix invalid free during queue allocation failure.

I've got a copy of the github mirror and wasnt sure how to convert the git commits back to SVN revisions, so I referenced the git commits where this was broken in my summary.

May 3 2018, 9:27 PM
jacob.e.keller_intel.com created D15285: iflib: fix invalid free during queue allocation failure.
May 3 2018, 9:26 PM
jacob.e.keller_intel.com created D15284: iflib: remove unused brscp pointer from iflib_queues_alloc.
May 3 2018, 9:25 PM

Aug 24 2017

jacob.e.keller_intel.com added a watcher for Intel Networking: jacob.e.keller_intel.com.
Aug 24 2017, 3:28 PM