mbox series

[v3,0/5] Fixes for dynamic boost control for 6.6

Message ID 20230829150759.156126-1-mario.limonciello@amd.com
Headers show
Series Fixes for dynamic boost control for 6.6 | expand

Message

Mario Limonciello Aug. 29, 2023, 3:07 p.m. UTC 
Testing the code that is submitted for kernel 6.6 found some occasional
problems with memory allocation and with the test suite and sample
application as merged.

This series fixes all problems found. Ideally all patches should
be submitted for a future 6.6-rc, but, the most important is
"crypto: ccp: Get a free page to use while fetching initial nonce"
if the rest must be pushed to 6.7.

This series merges together the two previous submissions, re-orders them
and fixes one more problem found with unfused tests.
Link: https://lore.kernel.org/linux-crypto/20230828190200.1598-1-mario.limonciello@amd.com/T/#m5f336a50707d2426c2a0d47d751f06d233e5c816
Link: https://lore.kernel.org/linux-crypto/20230824221932.2807-1-mario.limonciello@amd.com/#t

Mario Limonciello (5):
  crypto: ccp: Get a free page to use while fetching initial nonce
  crypto: ccp: Fix ioctl unit tests
  crypto: ccp: Fix DBC sample application error handling
  crypto: ccp: Fix sample application signature passing
  crypto: ccp: Fix some unfused tests

 drivers/crypto/ccp/dbc.c     |  2 +-
 tools/crypto/ccp/dbc.c       | 17 +++++++-------
 tools/crypto/ccp/dbc.py      |  8 +++----
 tools/crypto/ccp/test_dbc.py | 45 +++++++++++++++++++++---------------
 4 files changed, 41 insertions(+), 31 deletions(-)


base-commit: 85b9bf9a514d991fcecb118d0a8a35e754ff9265

Comments

Herbert Xu Aug. 31, 2023, 9:02 a.m. UTC | #1 
On Tue, Aug 29, 2023 at 10:07:54AM -0500, Mario Limonciello wrote:
>
> This series fixes all problems found. Ideally all patches should
> be submitted for a future 6.6-rc, but, the most important is
> "crypto: ccp: Get a free page to use while fetching initial nonce"
> if the rest must be pushed to 6.7.

As far as I can see it simply means that the new feature of DBC does
not work.  If that's all then I don't think it's an urgent issue and
this can wait until 6.7.

Cheers,
Mario Limonciello Aug. 31, 2023, 11:27 a.m. UTC | #2 
On 8/31/2023 04:02, Herbert Xu wrote:
> On Tue, Aug 29, 2023 at 10:07:54AM -0500, Mario Limonciello wrote:
>>
>> This series fixes all problems found. Ideally all patches should
>> be submitted for a future 6.6-rc, but, the most important is
>> "crypto: ccp: Get a free page to use while fetching initial nonce"
>> if the rest must be pushed to 6.7.
> 
> As far as I can see it simply means that the new feature of DBC does
> not work.  If that's all then I don't think it's an urgent issue and
> this can wait until 6.7.
> 
> Cheers,

It works; but it's a failure rate type issue based on what is in the 
page it gets.  I didn't reproduce it when it was initially submitted.
I only noticed it when testing from Linus' tree.
So I suspect it's related to other completely unrelated code that now
runs before ccp is loaded by the kernel.

So some boots you see the character device and can interact with it, 
others you can't.  If you start with 'ccp' loaded and it working, then 
unload/reload it might not come back.
Herbert Xu Aug. 31, 2023, 11:56 a.m. UTC | #3 
On Thu, Aug 31, 2023 at 06:27:59AM -0500, Mario Limonciello wrote:
>
> So some boots you see the character device and can interact with it, others
> you can't.  If you start with 'ccp' loaded and it working, then
> unload/reload it might not come back.

Is it just the new functionality that's broken or does it cause
existing functionality to fail as well?

Thanks,
Mario Limonciello Aug. 31, 2023, 4:02 p.m. UTC | #4 
On 8/31/2023 06:56, Herbert Xu wrote:
> On Thu, Aug 31, 2023 at 06:27:59AM -0500, Mario Limonciello wrote:
>>
>> So some boots you see the character device and can interact with it, others
>> you can't.  If you start with 'ccp' loaded and it working, then
>> unload/reload it might not come back.
> 
> Is it just the new functionality that's broken or does it cause
> existing functionality to fail as well?
> 
> Thanks,

Fortunately it's non-fatal for any other functionality in the driver.