Hi Boyan,
-----Original Message----- From: Boyan Karatotev boyan.karatotev@arm.com Sent: Wednesday, January 29, 2025 10:09 PM To: Bollapalli, Maheedhar Sai MaheedharSai.Bollapalli@amd.com; tf-a- tests@lists.trustedfirmware.org Subject: Re: [Tf-a-tests] Re: Invalid entry point in cpu hotplug
Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.
Hi Maheedhar,
There are other platforms which return SUCCESS for all address ranges in the
*_validate_ns_entrypoint API. Are these implementations incorrect ?
I suppose they are, yes. Presumably, if the supplied address on these platforms was out of range, we'd panic the core. It probably doesn't matter much in practice though as Linux won't usually do that.
Yes, there can be other platforms which will treat 0x0 as valid address. Considering this the tf-a-tests infrastructure for validating the entry point should be implemented in a way so that it can comply for all
platforms.
That's true, but it must not have came up this far with the existing platforms tftf supports. I don't see why it can't it be done, so if you have an idea for an implementation me or someone else on the team would be happy to review it.
To handle custom invalid entry addresses, we have updated a template change on one of our platforms to use the PLAT_INVALID_ADDR macro instead of the default 0x0. This change is part of the TFTF test for validating invalid entry addresses in the CPU hot plug scenario. A custom address can be provided either through the header (defaulting to the invalid address) or via the platform makefile to specify a custom valid or invalid address. Please review this idea and let us know your thoughts.
Please find the attached patch for the same.
Regards, Maheedhar.