linux-kernel - Re: [PATCH v1 1/2] cxl_test: enable zero sized decoders under hb0

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] [day] [month] [year] [list]

Message-ID: <aYynWqJ7u-v-6WsZ@gourry-fedora-PF4VCD3F>
Date: Wed, 11 Feb 2026 10:59:22 -0500
From: Gregory Price <gourry@...rry.net>
To: Alison Schofield <alison.schofield@...el.com>
Cc: Vishal Aslot <vaslot@...dia.com>, 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>,
	Peter Zijlstra <peterz@...radead.org>,
	"open list:COMPUTE EXPRESS LINK (CXL)" <linux-cxl@...r.kernel.org>,
	open list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v1 1/2] cxl_test: enable zero sized decoders under hb0

On Wed, Feb 11, 2026 at 10:04:41AM -0500, Gregory Price wrote:
> Revisiting this - it might be reasonable to allow post-lock ordering if
> *all* decoders in beyond the programmable one are zero-locked.
> 
> Seems ok? Thoughts?
> 

Sorry, got myself a bit mixed up, lets me a bit more precise, there are
annoying corner conditions here that make reasoning about this
difficult, and the mild ambiguity in the spec doesn't imbue confidence.

[open] = programmable
[programmed] = usually auto-decoder, but some cases may allow runtime
               programming, i haven't logic'd that out.


                    Post-lock   Legal
decoder        0             1             2     ...     N
------------------------------------------------------------------
            [open]       [zero-lock]  [zero-lock]   [zero-lock]
	    [open]         [open]     [zero-lock]   [zero-lock]
	  [programmed]     [open]     [zero-lock]   [zero-lock]
	  [programmed]  [programmed]  [zero-lock]   [zero-lock]


In these cases you basically act as-if the zero-locked decoders
essentially don't exist.



                   Pre-lock    Legal
decoder        0             1             2     ...     N
------------------------------------------------------------------
	  [zero-lock]    [zero-lock]  [zero-lock]      [open]
	  [zero-lock]    [zero-lock]  [zero-lock]    [programmed]
	  [zero-lock]    [zero-lock]  [programmed]     [open]
	  [zero-lock]    [zero-lock]  [programmed]   [programmed]

Again, in these causes you just act like the decoders don't exist.


                           Illegal
decoder        0              1             2      ...     N
------------------------------------------------------------------
            [open]       [zero-lock]      [open]      [zero-lock]
          [programmed]   [zero-lock]      [open]      [zero-lock]
            [open]       [zero-lock]   [programmed]   [zero-lock]
          [zero-lock]       [open]      [zero-lock]   [zero-lock]
          [zero-lock]       [open]      [zero-lock]     [open]
          [zero-lock]       [open]      [zero-lock]     [open]
          [zero-lock]       [open]      [zero-lock]   [programmed]

These cases are all illegal because there is always some form of
decoder committed out-of-order in the presence of at least one locked
decoder (the zero-locks).


                           Questionable
decoder        0              1             2      ...     N
------------------------------------------------------------------
	  [zero-lock]    [programmed]   [zero-lock]    [zero-lock]
          [programmed]   [zero-lock]    [programmed]   [zero-lock]
	  [zero-lock]    [programmed]   [zero-lock]    [programmed]
******    [zero-lock]    [programmed]   [zero-lock]  [open]...[open]


In these cases, it's only really legal IFF the programmed decoder(s) are
auto-decoders, because it means an out-of-order condition is impossible.

I think you can validate this on probe, but it's annoying

if zero-lock decoder is present
	a) all decoders prior to the zero-lock are locked (zero | auto)
	b) all decoders after the zero-lock are locked*

* if an open decoder appears after a zero-lock, all decoders after
  the open decoder must also be open. (supports corner case above)


I... think that works?  It's annoying though lol.

~Gregory