lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200127064007.GA12713@ogabbay-VM>
Date:   Mon, 27 Jan 2020 08:40:07 +0200
From:   Oded Gabbay <oded.gabbay@...il.com>
To:     gregkh@...uxfoundation.org
Cc:     linux-kernel@...r.kernel.org
Subject: [git pull] habanalabs pull request for kernel 5.6

Hello Greg,

This is the pull request for habanalabs driver for kernel 5.6.

It's a small one this time and mainly contains improvements to MMU and
reset code. Hopefully I will have time to work on upstreaming the Gaudi
support code for the next kernel, but it may be postponed to 5.8.

Please see the tag message for more details on what this pull request
contains.

In addition, I got my pgp key signed by Olof.J. I would appreciate it if
you could verify it.

Thanks,
Oded

The following changes since commit 0db4a15d4c2787b1112001790d4f95bd2c5fed6f:

  mei: me: add jasper point DID (2020-01-24 09:33:58 +0100)

are available in the Git repository at:

  git://people.freedesktop.org/~gabbayo/linux tags/misc-habanalabs-next-2020-01-27

for you to fetch changes up to 31579364e1478eda9556adc90168c11dc6bac150:

  habanalabs: patched cb equals user cb in device memset (2020-01-27 08:23:39 +0200)

----------------------------------------------------------------
This tag contains the following changes for kernel 5.6:

- MMU code improvements that includes:
  - Flush MMU TLB cache only once, at the end of mapping/unmapping
    function, instead of flushing after mapping of every page.
  - Add future ASIC support by splitting properties of ASIC capabilities
    regarding mapping of host memory to regular and huge pages.

- Fix bug where the driver didn't halt the internal ASIC's engines before
  doing hard-reset to the ASIC

- Minimize writes to H/W during hard-reset by skipping coresight disable
  code (no point as device is going to be hard-reset).

- Fix bug where we did an extra refcnt put in the memset device memory
  function.

- Small bug fixes and minor improvements to code.

----------------------------------------------------------------
Oded Gabbay (3):
      habanalabs: halt the engines before hard-reset
      habanalabs: removing extra ;
      habanalabs: patched cb equals user cb in device memset

Omer Shpigelman (3):
      habanalabs: use the user CB size as a default job size
      habanalabs: split the host MMU properties
      habanalabs: do not halt CoreSight during hard reset

Pawel Piskorski (1):
      habanalabs: flush only at the end of the map/unmap

Tomer Tayar (2):
      habanalabs: Modify CS jobs counter to u16
      habanalabs: Avoid running restore chunks if no execute chunks

 drivers/misc/habanalabs/command_submission.c  |  47 +++---
 drivers/misc/habanalabs/debugfs.c             |  21 ++-
 drivers/misc/habanalabs/device.c              |   7 +-
 drivers/misc/habanalabs/goya/goya.c           |  89 ++++++++---
 drivers/misc/habanalabs/goya/goya_coresight.c |   4 +-
 drivers/misc/habanalabs/habanalabs.h          |  39 ++---
 drivers/misc/habanalabs/memory.c              | 222 +++++++++++++++++---------
 drivers/misc/habanalabs/mmu.c                 | 110 ++++++++-----
 8 files changed, 346 insertions(+), 193 deletions(-)

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ