Linux CXL
 help / color / mirror / Atom feed
From: Gregory Price <gourry@gourry.net>
To: Robert Richter <rrichter@amd.com>
Cc: Alison Schofield <alison.schofield@intel.com>,
	Vishal Verma <vishal.l.verma@intel.com>,
	Ira Weiny <ira.weiny@intel.com>,
	Dan Williams <dan.j.williams@intel.com>,
	Jonathan Cameron <Jonathan.Cameron@huawei.com>,
	Dave Jiang <dave.jiang@intel.com>,
	Davidlohr Bueso <dave@stgolabs.net>,
	linux-cxl@vger.kernel.org, linux-kernel@vger.kernel.org,
	"Fabio M. De Francesco" <fabio.m.de.francesco@linux.intel.com>,
	Terry Bowman <terry.bowman@amd.com>
Subject: Re: [PATCH v2 04/15] cxl/region: Calculate endpoint's region position during init
Date: Fri, 4 Apr 2025 00:38:01 -0400	[thread overview]
Message-ID: <Z-9iKcmVz2Fj2q7Y@gourry-fedora-PF4VCD3F> (raw)
In-Reply-To: <20250218132356.1809075-5-rrichter@amd.com>

On Tue, Feb 18, 2025 at 02:23:45PM +0100, Robert Richter wrote:
> The calculation of an endpoint's position in a region traverses all
> ports up to the root port and determines the corresponding decoders
> for that particular address range. For address translation the HPA
> range must be recalculated between ports. In order to prepare the
> implementation of address translation, move code to
> cxl_endpoint_decoder_initialize() and reuse the existing iterator
> there.
> 
> Signed-off-by: Robert Richter <rrichter@amd.com>
> ---
>  drivers/cxl/core/region.c | 25 +++++++++++++++++++++----
>  1 file changed, 21 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/cxl/core/region.c b/drivers/cxl/core/region.c
> index ad4a6ce37216..6f106bfa115f 100644
> --- a/drivers/cxl/core/region.c
> +++ b/drivers/cxl/core/region.c
> @@ -1903,7 +1903,6 @@ static int cxl_region_sort_targets(struct cxl_region *cxlr)
>  	for (i = 0; i < p->nr_targets; i++) {
>  		struct cxl_endpoint_decoder *cxled = p->targets[i];
>  
> -		cxled->pos = cxl_calc_interleave_pos(cxled);
>  		/*

This change breaks the entire sort process, because in
cxl_region_attach_auto the positions are temporarily overwritten

To make this work, I had to drop over-write of the position when doing
the attach process - but this is just incorrect as we now have a
cxled->pos that indexes incorrectly into p->targets[N]

The result of the above change (without the below hack) is that decoder
probe order causes a race condition - whoever shows up first gets
position 0, and the sort does nothing (because the above line is
dropped).

TL;DR: This line should just be left as-is.  It's perfectly fine to
re-calculate the position.

~Gregory

--- DO NOT USE - added for context ---

diff --git a/drivers/cxl/core/region.c b/drivers/cxl/core/region.c
index 902b04b875b3..e75eb1c815f1 100644
--- a/drivers/cxl/core/region.c
+++ b/drivers/cxl/core/region.c
@@ -1855,7 +1855,6 @@ static int cxl_region_attach_auto(struct cxl_region *cxlr,
         */
        pos = p->nr_targets;
        p->targets[pos] = cxled;
-       cxled->pos = pos;
        p->nr_targets++;

        return 0;


  parent reply	other threads:[~2025-04-04  4:38 UTC|newest]

Thread overview: 61+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-02-18 13:23 [PATCH v2 00/15] cxl: Address translation support, part 2: Generic support and AMD Zen5 platform enablement Robert Richter
2025-02-18 13:23 ` [PATCH v2 01/15] cxl: Modify address translation callback for generic use Robert Richter
2025-02-20 16:00   ` Gregory Price
2025-02-20 21:03     ` Dave Jiang
2025-02-18 13:23 ` [PATCH v2 02/15] cxl: Introduce callback to translate an HPA range from a port to its parent Robert Richter
2025-02-20 21:19   ` Dave Jiang
2025-02-18 13:23 ` [PATCH v2 03/15] cxl/region: Factor out code for interleaving calculations Robert Richter
2025-02-20 16:28   ` Gregory Price
2025-02-20 16:41     ` Gregory Price
2025-03-14 12:45   ` Jonathan Cameron
2025-02-18 13:23 ` [PATCH v2 04/15] cxl/region: Calculate endpoint's region position during init Robert Richter
2025-02-19 23:32   ` Gregory Price
2025-02-20 17:31   ` Gregory Price
2025-02-20 21:56   ` Dave Jiang
2025-04-04  4:38   ` Gregory Price [this message]
2025-04-04 15:36   ` [PATCH] cxl/region: Continue recalculating position during sort Gregory Price
2025-04-04 17:22     ` Gregory Price
2025-04-05  2:35   ` [PATCH] cxl region: recalculate interleave pos during region probe Gregory Price
2025-04-08 15:30     ` Gregory Price
2025-02-18 13:23 ` [PATCH v2 05/15] cxl/region: Calculate and store the SPA range of an endpoint Robert Richter
2025-02-20 18:42   ` Gregory Price
2025-02-20 22:31   ` Dave Jiang
2025-02-20 22:37     ` Gregory Price
2025-03-14 12:41   ` Jonathan Cameron
2025-02-18 13:23 ` [PATCH v2 06/15] cxl/region: Use endpoint's HPA range to find the port's decoder Robert Richter
2025-04-24  0:28   ` Gregory Price
2025-04-24 21:49     ` Gregory Price
2025-04-24 23:46       ` Gregory Price
2025-02-18 13:23 ` [PATCH v2 07/15] cxl/region: Use translated HPA ranges " Robert Richter
2025-02-20 19:13   ` Gregory Price
2025-02-18 13:23 ` [PATCH v2 08/15] cxl/region: Use the endpoint's SPA range to find a region Robert Richter
2025-02-20 19:28   ` Gregory Price
2025-03-14 12:45   ` Jonathan Cameron
2025-04-08 15:45   ` Gregory Price
2025-02-18 13:23 ` [PATCH v2 09/15] cxl/region: Use the endpoint's SPA range to create " Robert Richter
2025-02-20 19:31   ` Gregory Price
2025-03-14 12:46   ` Jonathan Cameron
2025-04-08 15:50   ` Gregory Price
2025-02-18 13:23 ` [PATCH v2 10/15] cxl/region: Use root decoders interleaving parameters " Robert Richter
2025-02-20 19:43   ` Gregory Price
2025-03-14 12:49   ` Jonathan Cameron
2025-04-01  1:59   ` Gregory Price
2025-04-01  5:26     ` Gregory Price
2025-04-01 18:03   ` Gregory Price
2025-02-18 13:23 ` [PATCH v2 11/15] cxl/region: Use the endpoint's SPA range to check " Robert Richter
2025-02-20 19:50   ` Gregory Price
2025-04-08 15:54   ` Gregory Price
2025-02-18 13:23 ` [PATCH v2 12/15] cxl/region: Lock decoders that need address translation Robert Richter
2025-02-20 19:57   ` Gregory Price
2025-02-18 13:23 ` [PATCH v2 13/15] cxl/x86: Prepare for architectural platform setup Robert Richter
2025-02-20 19:57   ` Gregory Price
2025-02-18 13:23 ` [PATCH v2 14/15] cxl/amd: Enable Zen5 address translation using ACPI PRMT Robert Richter
2025-02-21  0:40   ` Dave Jiang
2025-03-14 13:01   ` Jonathan Cameron
2025-04-05  2:38   ` [PATCH] [HACK] drop zen5_init checks due to segfault Gregory Price
2025-05-13 21:10     ` Robert Richter
2025-06-17 20:33       ` Joshua Hahn
2025-06-24  5:43         ` Robert Richter
2025-06-24 21:46           ` Joshua Hahn
2025-02-18 13:23 ` [PATCH v2 15/15] MAINTAINERS: CXL: Add entry for AMD platform support (CXL_AMD) Robert Richter
2025-02-20  1:00 ` [PATCH v2 00/15] cxl: Address translation support, part 2: Generic support and AMD Zen5 platform enablement Gregory Price

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=Z-9iKcmVz2Fj2q7Y@gourry-fedora-PF4VCD3F \
    --to=gourry@gourry.net \
    --cc=Jonathan.Cameron@huawei.com \
    --cc=alison.schofield@intel.com \
    --cc=dan.j.williams@intel.com \
    --cc=dave.jiang@intel.com \
    --cc=dave@stgolabs.net \
    --cc=fabio.m.de.francesco@linux.intel.com \
    --cc=ira.weiny@intel.com \
    --cc=linux-cxl@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=rrichter@amd.com \
    --cc=terry.bowman@amd.com \
    --cc=vishal.l.verma@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox