From: Zijun Hu <zijun_hu@icloud.com>
To: Dan Williams <dan.j.williams@intel.com>,
quic_zijuhu <quic_zijuhu@quicinc.com>,
Ira Weiny <ira.weiny@intel.com>,
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Davidlohr Bueso <dave@stgolabs.net>,
Jonathan Cameron <jonathan.cameron@huawei.com>,
Dave Jiang <dave.jiang@intel.com>,
Alison Schofield <alison.schofield@intel.com>,
Vishal Verma <vishal.l.verma@intel.com>,
Timur Tabi <timur@kernel.org>,
"David S. Miller" <davem@davemloft.net>,
Eric Dumazet <edumazet@google.com>,
Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
linux-cxl@vger.kernel.org, linux-kernel@vger.kernel.org,
netdev@vger.kernel.org
Subject: Re: [PATCH v4 1/2] cxl/region: Find free cxl decoder by device_for_each_child()
Date: Wed, 11 Sep 2024 20:14:48 +0800 [thread overview]
Message-ID: <11576596-f0e4-4e88-a200-ed22b86d5974@icloud.com> (raw)
In-Reply-To: <66e08f9beb6a2_326462945d@dwillia2-xfh.jf.intel.com.notmuch>
On 2024/9/11 02:27, Dan Williams wrote:
> Dan Williams wrote:
> [..]
>> So, while regionB would be the next decoder to allocate after regionC is
>> torn down, it is not a free decoder while decoderC and regionC have not been
>> reclaimed.
>
> The "simple" conversion is bug compatible with the current
> implementation, but here's a path to both constify the
> device_find_child() argument, *and* prevent unwanted allocations by
> allocating decoders precisely by id. Something like this, it passes a
> quick unit test run:
>
sounds good.
> diff --git a/drivers/cxl/core/port.c b/drivers/cxl/core/port.c
> index 1d5007e3795a..749a281819b4 100644
> --- a/drivers/cxl/core/port.c
> +++ b/drivers/cxl/core/port.c
> @@ -1750,7 +1750,8 @@ static int cxl_decoder_init(struct cxl_port *port, struct cxl_decoder *cxld)
> struct device *dev;
> int rc;
>
> - rc = ida_alloc(&port->decoder_ida, GFP_KERNEL);
> + rc = ida_alloc_max(&port->decoder_ida, CXL_DECODER_NR_MAX - 1,
> + GFP_KERNEL);
> if (rc < 0)
> return rc;
>
> diff --git a/drivers/cxl/core/region.c b/drivers/cxl/core/region.c
> index 21ad5f242875..1f7b3a9ebfa3 100644
> --- a/drivers/cxl/core/region.c
> +++ b/drivers/cxl/core/region.c
> @@ -794,26 +794,16 @@ static size_t show_targetN(struct cxl_region *cxlr, char *buf, int pos)
> return rc;
> }
>
> -static int match_free_decoder(struct device *dev, void *data)
> +static int match_decoder_id(struct device *dev, void *data)
> {
> struct cxl_decoder *cxld;
> - int *id = data;
> + int id = *(int *) data;
>
> if (!is_switch_decoder(dev))
> return 0;
>
> cxld = to_cxl_decoder(dev);
> -
> - /* enforce ordered allocation */
> - if (cxld->id != *id)
> - return 0;
> -
> - if (!cxld->region)
> - return 1;
> -
> - (*id)++;
> -
> - return 0;
> + return cxld->id == id;
> }
>
> static int match_auto_decoder(struct device *dev, void *data)
> @@ -840,16 +830,29 @@ cxl_region_find_decoder(struct cxl_port *port,
> struct cxl_region *cxlr)
> {
> struct device *dev;
> - int id = 0;
> -
> if (port == cxled_to_port(cxled))
> return &cxled->cxld;
>
> if (test_bit(CXL_REGION_F_AUTO, &cxlr->flags))
> dev = device_find_child(&port->dev, &cxlr->params,
> match_auto_decoder);
> - else
> - dev = device_find_child(&port->dev, &id, match_free_decoder);
> + else {
> + int id, last;
> +
> + /*
> + * Find next available decoder, but fail new decoder
> + * allocations if out-of-order region destruction has
> + * occurred
> + */
> + id = find_first_zero_bit(port->decoder_alloc,
> + CXL_DECODER_NR_MAX);
> + last = find_last_bit(port->decoder_alloc, CXL_DECODER_NR_MAX);
> +
> + if (id >= CXL_DECODER_NR_MAX ||
> + (last < CXL_DECODER_NR_MAX && id != last + 1))
> + return NULL;
Above finding logic seems wrong.
what about below one ?
last = find_last_bit(port->decoder_alloc, CXL_DECODER_NR_MAX);
if (last >= CXL_DECODER_NR_MAX)
id = 0;
else if (last + 1 < CXL_DECODER_NR_MAX)
id = last + 1;
else
return NULL;
> + dev = device_find_child(&port->dev, &id, match_decoder_id);
> + }
> if (!dev)
> return NULL;
> /*
> @@ -943,6 +946,9 @@ static void cxl_rr_free_decoder(struct cxl_region_ref *cxl_rr)
>
> dev_WARN_ONCE(&cxlr->dev, cxld->region != cxlr, "region mismatch\n");
> if (cxld->region == cxlr) {
> + struct cxl_port *port = to_cxl_port(cxld->dev.parent);
> +
> + clear_bit(cxld->id, port->decoder_alloc);
> cxld->region = NULL;
> put_device(&cxlr->dev);
> }
> @@ -977,6 +983,7 @@ static int cxl_rr_ep_add(struct cxl_region_ref *cxl_rr,
> cxl_rr->nr_eps++;
>
> if (!cxld->region) {
> + set_bit(cxld->id, port->decoder_alloc);
> cxld->region = cxlr;
> get_device(&cxlr->dev);
> }
> diff --git a/drivers/cxl/cxl.h b/drivers/cxl/cxl.h
> index 9afb407d438f..750cd027d0b0 100644
> --- a/drivers/cxl/cxl.h
> +++ b/drivers/cxl/cxl.h
> @@ -578,6 +578,9 @@ struct cxl_dax_region {
> struct range hpa_range;
> };
>
> +/* Max as of CXL 3.1 (8.2.4.20.1 CXL HDM Decoder Capability Register) */
> +#define CXL_DECODER_NR_MAX 32
> +
> /**
> * struct cxl_port - logical collection of upstream port devices and
> * downstream port devices to construct a CXL memory
> @@ -591,6 +594,7 @@ struct cxl_dax_region {
> * @regions: cxl_region_ref instances, regions mapped by this port
> * @parent_dport: dport that points to this port in the parent
> * @decoder_ida: allocator for decoder ids
> + * @decoder_alloc: decoder busy/free (@cxld->region set) bitmap
> * @reg_map: component and ras register mapping parameters
> * @nr_dports: number of entries in @dports
> * @hdm_end: track last allocated HDM decoder instance for allocation ordering
> @@ -611,6 +615,7 @@ struct cxl_port {
> struct xarray regions;
> struct cxl_dport *parent_dport;
> struct ida decoder_ida;
> + DECLARE_BITMAP(decoder_alloc, CXL_DECODER_NR_MAX);
> struct cxl_register_map reg_map;
> int nr_dports;
> int hdm_end;
next prev parent reply other threads:[~2024-09-11 12:15 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-09-05 0:36 [PATCH v4 0/2] driver core: Prevent device_find_child() from modifying caller's match data Zijun Hu
2024-09-05 0:36 ` [PATCH v4 1/2] cxl/region: Find free cxl decoder by device_for_each_child() Zijun Hu
2024-09-05 5:32 ` Greg Kroah-Hartman
2024-09-05 8:48 ` quic_zijuhu
2024-09-05 11:18 ` Zijun Hu
2024-09-09 19:56 ` Ira Weiny
2024-09-10 0:45 ` Dan Williams
2024-09-10 3:17 ` quic_zijuhu
2024-09-10 4:15 ` Dan Williams
2024-09-10 4:20 ` Dan Williams
2024-09-10 11:46 ` Zijun Hu
2024-09-10 16:01 ` Dan Williams
2024-09-10 18:27 ` Dan Williams
2024-09-11 12:14 ` Zijun Hu [this message]
2024-09-11 11:52 ` Zijun Hu
2024-09-05 0:36 ` [PATCH v4 2/2] net: qcom/emac: Find sgmii_ops " Zijun Hu
2024-09-05 5:29 ` Greg Kroah-Hartman
2024-09-05 5:33 ` Greg Kroah-Hartman
2024-09-05 9:09 ` quic_zijuhu
2024-09-06 0:29 ` Zijun Hu
2024-09-05 8:29 ` quic_zijuhu
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=11576596-f0e4-4e88-a200-ed22b86d5974@icloud.com \
--to=zijun_hu@icloud.com \
--cc=alison.schofield@intel.com \
--cc=dan.j.williams@intel.com \
--cc=dave.jiang@intel.com \
--cc=dave@stgolabs.net \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=gregkh@linuxfoundation.org \
--cc=ira.weiny@intel.com \
--cc=jonathan.cameron@huawei.com \
--cc=kuba@kernel.org \
--cc=linux-cxl@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=netdev@vger.kernel.org \
--cc=pabeni@redhat.com \
--cc=quic_zijuhu@quicinc.com \
--cc=timur@kernel.org \
--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