PCIe device ID and vendor ID fields are swapped in CSL register maps

Paul Barker @ Commagility

In both C6678 PDK and Keystone 2 PDK, the constants for the PCI VENDOR_DEVICE_ID registers are wrong. This affects the usage of both the chip support library (CSL) and the PCIe low-level driver (LLD).

This has been reported before: https://e2e.ti.com/support/dsp/c6000_multi-core_dsps/f/639/t/385456 and https://e2e.ti.com/support/dsp/c6000_multi-core_dsps/f/639/p/321624/1119146. However, no fix has been confirmed. This issue is present in the latest MCSDK release for Keystone 2 (3.01.01.04).

Please advise whether this will be fixed in the next MCSDK release.

over 9 years ago

0 Raja over 9 years ago

TI__Guru* 81335 points

Hi Paul,
Thank you for the post.
We will ping the factory team again for this issue. The CSL macros values needs to be updated.

0 Ganapathi Dhandapani95 over 9 years ago

TI__Mastermind 28085 points

Hi,

These CSL macros values are corrected in MCSDK 2.1.3.x package. But i don't know the exact timeline for release of this MCSDK version.

Thanks,

0 Raja over 9 years ago in reply to Ganapathi Dhandapani95

TI__Guru* 81335 points

Ganapathi,
Thank you for checking this one.

0 Paul Barker @ Commagility over 9 years ago in reply to Raja

Expert 1490 points

I'm a bit confused as the latest release of MCSDK is 3.01.03.06. I've compared the files "packages/ti/csl/cslr_pcie_cfg_space_endpoint.h" between release 3.01.01.04 and 3.01.03.06 and they're bit-for-bit identical so the defines haven't been fixed. Could you look into this again for me and let me know when it will be fixed in MCSDK for Keystone II?

Thanks,
Paul Barker

0 Ganapathi Dhandapani95 over 9 years ago in reply to Paul Barker @ Commagility

TI__Mastermind 28085 points

Hi,

That is Keystone I MCSDK 2.1.3.x package. But you compared Keystone II MCSDK packages.

Thanks,

0 Paul Barker @ Commagility over 9 years ago in reply to Ganapathi Dhandapani95

Expert 1490 points

Yes, I'm asking when it will be fixed for both the Keystone I MCSDK and the Keystone II MCSDK.

By my understanding: the latest release of Keystone I MCSDK is 2.01.02.06, released 28th Feb 2013. The latest releast of Keystone II MCSDK is 3.01.03.06, released 27th March 2015. Both of these have the issue I've described. This issue was reported over a year ago (e2e.ti.com/.../1119146) and hasn't been fixed yet.

Regards,
Paul Barker

0 Ganapathi Dhandapani95 over 9 years ago in reply to Paul Barker @ Commagility

TI__Mastermind 28085 points

Hi,

I have filed bug report for Keystone II MCSDK, It may get resolved in the future release. Thank you for your feedback.

It is a CSL library issue, DSP raw register shows a correct PCIe VENDOR and DEVICE ID values.

Thanks,

0 Paul Barker @ Commagility over 9 years ago in reply to Ganapathi Dhandapani95

Expert 1490 points

Ganapathi,

Is it possible to target this fix for the next release of the Keystone I and Keystone II MCSDK packages? This may become a customer facing issue for us as we have customers planning to communicate between our systems using PCIe. The fix itself should be very simple, the device ID and vendor ID mask and shift constants just need to be swapped around.

Many thanks for your help,
Paul

0 Ganapathi Dhandapani95 over 9 years ago in reply to Paul Barker @ Commagility

TI__Mastermind 28085 points

It may get resolved in the future Keystone II MCSDK release. Better to suggest the simple workaround to your customers.

Thanks,

0 Paul Barker @ Commagility over 9 years ago in reply to Ganapathi Dhandapani95

Expert 1490 points

Ganapathi,

Thanks for looking into this. I hope that this can be resolved in the next MCSDK release.

Paul

Processors

Processors forum

PCIe device ID and vendor ID fields are swapped in CSL register maps