2017-06-19 18:42:57

by Jon Mason

[permalink] [raw]
Subject: [GIT PULL] NTB bug fixes for v4.12

Hello Linus,
Here are a few NTB bug fixes for 4.12. Please consider pulling them.

Thanks,
Jon



The following changes since commit 41f1830f5a7af77cf5c86359aba3cbd706687e52:

Linux 4.12-rc6 (2017-06-19 22:19:37 +0800)

are available in the git repository at:

git://github.com/jonmason/ntb tags/ntb-4.12-bugfixes

for you to fetch changes up to 88931ec3dc11e7dbceb3b0df455693873b508fbe:

ntb: no sleep in ntb_async_tx_submit (2017-06-19 14:24:41 -0400)

----------------------------------------------------------------
NTB bug fixes to address the modinfo in ntb_perf, a couple of bugs in
the NTB transport QP calculations, skx doorbells, and sleeping in
ntb_async_tx_submit.

----------------------------------------------------------------
Allen Hubbe (1):
ntb: no sleep in ntb_async_tx_submit

Dave Jiang (1):
ntb: ntb_hw_intel: Skylake doorbells should be 32bits, not 64bits

Gary R Hook (1):
ntb: Correct modinfo usage statement for ntb_perf

Logan Gunthorpe (3):
NTB: ntb_test: fix bug printing ntb_perf results
ntb_transport: fix qp count bug
ntb_transport: fix bug calculating num_qps_mw

drivers/ntb/hw/intel/ntb_hw_intel.c | 2 +-
drivers/ntb/ntb_transport.c | 58 +++++++--------------------------
drivers/ntb/test/ntb_perf.c | 4 +--
tools/testing/selftests/ntb/ntb_test.sh | 2 +-
4 files changed, 15 insertions(+), 51 deletions(-)


2017-06-20 03:06:22

by Linus Torvalds

[permalink] [raw]
Subject: Re: [GIT PULL] NTB bug fixes for v4.12

On Tue, Jun 20, 2017 at 2:42 AM, Jon Mason <[email protected]> wrote:
> Hello Linus,
> Here are a few NTB bug fixes for 4.12.

So I pulled this, mainly because it removed more lines than it added
due to the revert.

But generally I absolutely *hate* pulling stuff that I can see was
committed just hours ago. Why was that branch so recent? It very
obviously cannot have been in any linux-next or 0day or gotten any
other exposure. Is there a reason for that behavior? And if there is,
just include it in the "please pull", because if it wasn't for the
"removes more lines than adds" I probably would just have decided to
ignore this.

Linus

2017-06-20 14:59:51

by Jon Mason

[permalink] [raw]
Subject: Re: [GIT PULL] NTB bug fixes for v4.12

On Mon, Jun 19, 2017 at 11:06 PM, Linus Torvalds
<[email protected]> wrote:
> On Tue, Jun 20, 2017 at 2:42 AM, Jon Mason <[email protected]> wrote:
>> Hello Linus,
>> Here are a few NTB bug fixes for 4.12.
>
> So I pulled this, mainly because it removed more lines than it added
> due to the revert.
>
> But generally I absolutely *hate* pulling stuff that I can see was
> committed just hours ago. Why was that branch so recent? It very
> obviously cannot have been in any linux-next or 0day or gotten any
> other exposure. Is there a reason for that behavior? And if there is,

This was in my ntb branch since June 9th. Previously, it was based on
v4.12-rc4. Immediately prior to doing the pull request, I did a
rebase to rc6 to avoid any potential issues you might have applying
them.

> just include it in the "please pull", because if it wasn't for the
> "removes more lines than adds" I probably would just have decided to
> ignore this.

In the future, I will endeavor to be more verbose. I will include the
previous version that it was based on and the testing done by the NTB
developer community. I apologize for not doing this in this pull
request, and I appreciate you asking me and not simply ignoring my
email.

Thanks,
Jon

>
> Linus

2017-06-21 19:05:58

by Linus Torvalds

[permalink] [raw]
Subject: Re: [GIT PULL] NTB bug fixes for v4.12

On Tue, Jun 20, 2017 at 7:59 AM, Jon Mason <[email protected]> wrote:
>
> This was in my ntb branch since June 9th. Previously, it was based on
> v4.12-rc4. Immediately prior to doing the pull request, I did a
> rebase to rc6 to avoid any potential issues you might have applying
> them.

Please don't do that.

There are valid reasons for rebasing, but "just because" is not one of
them. The valid reasons are things like "I noticed that a completely
broken patch had made it into my queue, so I rebased to remove it" or
similar.

One common theme for valid reasons tends to be "..and I needed to
rerun all the testing", so in general I get very suspicious if things
are very recent, because clearly they didn't get very well tested.

Linus