netdev - Re: [PATCH v9 03/27] cxl: add capabilities field to cxl_dev_state and cxl

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-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <67902229a7e7f_20fa29437@dwillia2-xfh.jf.intel.com.notmuch>
Date: Tue, 21 Jan 2025 14:39:37 -0800
From: Dan Williams <dan.j.williams@...el.com>
To: Alejandro Lucero Palau <alucerop@....com>, Dan Williams
	<dan.j.williams@...el.com>, <alejandro.lucero-palau@....com>,
	<linux-cxl@...r.kernel.org>, <netdev@...r.kernel.org>, <edward.cree@....com>,
	<davem@...emloft.net>, <kuba@...nel.org>, <pabeni@...hat.com>,
	<edumazet@...gle.com>, <dave.jiang@...el.com>
Subject: Re: [PATCH v9 03/27] cxl: add capabilities field to cxl_dev_state
 and cxl_port

Alejandro Lucero Palau wrote:
[..]
> > I do not understand the rationale for a capability bitmap. There is
> > already a 'valid' flag in 'struct cxl_reg_map' for all register blocks.
> > Any optional core functionality should key off those existing flags.
> 
> 
> The current code is based on Type3 and the registers and capabilities 
> are defined as mandatory, I think except RAS.
> 
> With Type2 we have optional capabilities like mailbox and hdm, and the 
> code probing the regs should not make any assumption about what should 
> be there.
> 
> With this patchset the capabilities to expect are set by the accel 
> driver and compared with those discovered when probing CXL regs. 
> Although the capabilities check could use the cxl_reg_map, I consider it 
> is convenient to have a capability bitmap for keeping those discovered 
> and easily checking them against those expected by the accel driver, and 
> reporting them (if necessary) as well without further processing.

That is just on-loading redundancy to the core data structure for a
workalike way of checking the available register blocks. The 'struct
cxl_reg_map' already tracks this, the only needed change is to move the
responsibility for validating those bits to the driver. That work is
nearly identical to teaching the driver to inject a capability bitmask,
but more flexible for the case where the driver wants to optionally
enable functionality. In that latter case it will end up checking the
valid-bits anyway.