Adds new tap_delays data for SPM.
From internal git commit:
1f74169475d9023754c35ea150933e7eed67da94
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Josh Boyer <jwboyer@kernel.org>
Adds new tap_delays data for SPM.
From internal git commit:
1f74169475d9023754c35ea150933e7eed67da94
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Josh Boyer <jwboyer@kernel.org>
Adds new tap_delays data for SPM.
From internal git commit:
1f74169475d9023754c35ea150933e7eed67da94
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Josh Boyer <jwboyer@kernel.org>
Adds new tap_delays data for SPM.
From internal git commit:
1f74169475d9023754c35ea150933e7eed67da94
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Josh Boyer <jwboyer@kernel.org>
Adds new tap_delays data for SPM.
From internal git commit:
1f74169475d9023754c35ea150933e7eed67da94
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Josh Boyer <jwboyer@kernel.org>
Qualcomm binary files can be published in one of two formats: a split
MDT+Bnn format, consisting of several files, or a squashed MBN format,
consisting of a single ELF file.
Most of Qualcomm binary files (except the venus firmware) are
distributed (in linux-firmware) and used in the MBN format (even if the
original publishing format was MDT), since that requires transfer of a
single file between the userspace and the kernel.
Provide a squashed variant of Qualcomm Adreno a530 ZAP shader. For
backwards compatibility reason the split a530_zap.* files are retained.
They can be removed later, after all users migrate to new file location.
Cc: Rob Clark <robdclark@chromium.org>
Cc: Bjorn Andersson <andersson@kernel.org>
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Direction from upstream is to use minimal version numbering on
firmware files and replace rather than add. So add versionless
editions of the HuC files for currently HuC enabled platforms.
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Direction from upstream is to use minimal version numbering on
firmware files and replace rather than add. So going forwards, GuC
files will use only the major version number and HuC files will be
completely versionless.
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
After some discussion with Bjorn, the Qualcomm platform maintainer, it
was agreed to follow the established practice and to inject the SoC name
into the firmware paths.
It is highly unlikely that we are going to run Linux off the existing
ACPI data in these platforms anytime soon. And even if we do so, we'd
have to figure out the file names anyway, as they do not follow the
existing standard names.
Cc: Bjorn Andersson <andersson@kernel.org>
Cc: Mark Pearson <markpearson@lenovo.com>
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Acked-by: Bjorn Andersson <andersson@kernel.org>
Signed-off-by: Josh Boyer <jwboyer@kernel.org>
Add latest verified version of Nvidia Spectrum-family switch firmware,
for Spectrum (13.2010.3146), Spectrum-2 (29.2010.3146) and Spectrum-3
(30.2010.3146). Furthermore, add lc_ini_bundle_2010_3146.bin for modular
systems.
The following issues have been fixed:
- In Spectrum-3, fix fallback from PCIe gen3 to gen1, resulting in poor PCI
throughput
- In Spectrum-2, 1Gbps speed is now supported with 40Gbps modules
- Stability issues when burning FW to gearboxes on modular systems
- Several issues in link negotiation and link stability of modular systems
- Latency of several registers has been improved: PLLP, SFMR, SFN, PEFA,
RATR, MPILM, PPBS, RMPE, RIPS, PTER, SBCTR
- In Spectrum-2 and Spectrum-3, an issue in ECN marking on traffic going
through split ports
- Support of IPv4-to-IPv6 NAT in Spectrum-4
- PTP accuracy improvements on 1Gbps, 25Gbps and 50Gbps ports in Spectrum-1
and 10Gbps port in Spectrum-3
- On Spectrum-2+, fix edge cases of ACL matching IPv6 packets
- On Spectrum-3, an issue with a partial CRC overwrite when timestamp over
CRC is set to UTC and CRC recalculation is disabled
- On Spectrum-1, a possible FW freeze when many (e.g. 70) ports are up, and
shared buffer configuration is applied on the fly
- On Spectrum-3, an issue with port-down reason still being reported
despite link having been established
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Josh Boyer <jwboyer@kernel.org>
Since the licence information in the WHENCE for MediaTek firmwares
are identical to the LICENCE.mediatek, clean up the WHENCE to
reference the licence file.
Signed-off-by: Tinghan Shen <tinghan.shen@mediatek.com>
