linux-kernel - Re: [PATCH v1 2/2] cxl: Allow zero sized HDM decoders

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: <aPXazKH-FYS5K3qq@aschofie-mobl2.lan>
Date: Sun, 19 Oct 2025 23:46:36 -0700
From: Alison Schofield <alison.schofield@...el.com>
To: Vishal Aslot <vaslot@...dia.com>
CC: Davidlohr Bueso <dave@...olabs.net>, Jonathan Cameron
	<jonathan.cameron@...wei.com>, Dave Jiang <dave.jiang@...el.com>, "Vishal
 Verma" <vishal.l.verma@...el.com>, Ira Weiny <ira.weiny@...el.com>, "Dan
 Williams" <dan.j.williams@...el.com>, Li Ming <ming.li@...omail.com>, "open
 list:COMPUTE EXPRESS LINK (CXL)" <linux-cxl@...r.kernel.org>, open list
	<linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v1 2/2] cxl: Allow zero sized HDM decoders

On Tue, Oct 14, 2025 at 07:40:06PM -0700, Vishal Aslot wrote:
> CXL spec permits committing zero sized decoders.
> Linux currently considers them as an error.
> 
> Zero-sized decoders are helpful when the BIOS
> is committing them. Often BIOS will also lock
> them to prevent them being changed due to the
> TSP requirement. For example, if the type 3
> device is part of a TCB.
> 
> The host bridge, switch, and end-point decoders
> can all be committed with zero-size. If they are
> locked along the VH, it is often to prevent
> hotplugging of a new device that could not be
> attested post boot and cannot be included in
> TCB.
> 
> The caller leaves the decoder allocated but does
> not add it. It simply continues to the next decoder.

Please write out acronyms TSP and TCB ?
Commit log wraps at 70.

I tried out the cxl-test patch of this. And came up with the same
question that DaveJ responded here.  Why aren't we adding these
decoders to the topology? Are these guaranteed to always be lesser
id's than than the decoders for any BIOS defined regions, or is
the driver expected to handle non contiguous decoder sets?

It's an accounting task, whether they are in the topology or not,
but guessing in the topology may make that accounting easier.


> 
> Signed-off-by: Vishal Aslot <vaslot@...dia.com>
> ---
>  drivers/cxl/core/hdm.c | 9 ++++++---
>  1 file changed, 6 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/cxl/core/hdm.c b/drivers/cxl/core/hdm.c
> index d3a094ca01ad..1c036a485723 100644
> --- a/drivers/cxl/core/hdm.c
> +++ b/drivers/cxl/core/hdm.c
> @@ -1036,13 +1036,14 @@ static int init_hdm_decoder(struct cxl_port *port, struct cxl_decoder *cxld,
>  			return -ENXIO;
>  		}
>  
> +		port->commit_end = cxld->id;
> +
>  		if (size == 0) {
> -			dev_warn(&port->dev,
> +			dev_dbg(&port->dev,
>  				 "decoder%d.%d: Committed with zero size\n",
>  				 port->id, cxld->id);
> -			return -ENXIO;
> +			return -ENOSPC;
>  		}
> -		port->commit_end = cxld->id;
>  	} else {
>  		if (cxled) {
>  			struct cxl_memdev *cxlmd = cxled_to_memdev(cxled);
> @@ -1198,6 +1199,8 @@ static int devm_cxl_enumerate_decoders(struct cxl_hdm *cxlhdm,
>  
>  		rc = init_hdm_decoder(port, cxld, hdm, i, &dpa_base, info);
>  		if (rc) {
> +			if (rc == -ENOSPC)
> +				continue;

This needs to be in the mock_devm_cxl_enumerate_decoders() ....

>  			dev_warn(&port->dev,
>  				 "Failed to initialize decoder%d.%d\n",
>  				 port->id, i);
> -- 
> 2.43.0
>