From: Jonathan Cameron <Jonathan.Cameron@Huawei.com>
To: Lukas Wunner <lukas@wunner.de>
Cc: Bjorn Helgaas <helgaas@kernel.org>, <linux-pci@vger.kernel.org>,
"Gregory Price" <gregory.price@memverge.com>,
Ira Weiny <ira.weiny@intel.com>,
"Dan Williams" <dan.j.williams@intel.com>,
Alison Schofield <alison.schofield@intel.com>,
Vishal Verma <vishal.l.verma@intel.com>,
Dave Jiang <dave.jiang@intel.com>,
"Li, Ming" <ming4.li@intel.com>,
"Hillf Danton" <hdanton@sina.com>,
Ben Widawsky <bwidawsk@kernel.org>, <linuxarm@huawei.com>,
<linux-cxl@vger.kernel.org>
Subject: Re: [PATCH v3 16/16] cxl/pci: Rightsize CDAT response allocation
Date: Tue, 14 Feb 2023 13:05:52 +0000 [thread overview]
Message-ID: <20230214130552.000030ea@Huawei.com> (raw)
In-Reply-To: <49c5299afc660ac33fee9a116ea37df0de938432.1676043318.git.lukas@wunner.de>
On Fri, 10 Feb 2023 21:25:16 +0100
Lukas Wunner <lukas@wunner.de> wrote:
> Jonathan notes that cxl_cdat_get_length() and cxl_cdat_read_table()
> allocate 32 dwords for the DOE response even though it may be smaller.
>
> In the case of cxl_cdat_get_length(), only the second dword of the
> response is of interest (it contains the length). So reduce the
> allocation to 2 dwords and let DOE discard the remainder.
>
> In the case of cxl_cdat_read_table(), a correctly sized allocation for
> the full CDAT already exists. Let DOE write each table entry directly
> into that allocation. There's a snag in that the table entry is
> preceded by a Table Access Response Header (1 dword). Save the last
> dword of the previous table entry, let DOE overwrite it with the
> header of the next entry and restore it afterwards.
Marginally nasty, but looks like it works to me and avoid excessive allocations.
>
> The resulting CDAT is preceded by 4 unavoidable useless bytes. Increase
> the allocation size accordingly and skip these bytes when exposing CDAT
> in sysfs.
>
> The buffer overflow check in cxl_cdat_read_table() becomes unnecessary
> because the remaining bytes in the allocation are tracked in "length",
> which is passed to DOE and limits how many bytes it writes to the
> allocation. Additionally, cxl_cdat_read_table() bails out if the DOE
> response is truncated due to insufficient space.
>
> Tested-by: Ira Weiny <ira.weiny@intel.com>
> Signed-off-by: Lukas Wunner <lukas@wunner.de>
> Cc: Jonathan Cameron <Jonathan.Cameron@Huawei.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Follow up comment on earlier one inline.
> ---
> Changes v2 -> v3:
> * Newly added patch in v3 on popular request (Jonathan)
>
> drivers/cxl/core/pci.c | 34 ++++++++++++++++++----------------
> drivers/cxl/cxl.h | 3 ++-
> drivers/cxl/port.c | 2 +-
> 3 files changed, 21 insertions(+), 18 deletions(-)
>
> diff --git a/drivers/cxl/core/pci.c b/drivers/cxl/core/pci.c
> index 1b954783b516..70097cc75302 100644
> --- a/drivers/cxl/core/pci.c
> +++ b/drivers/cxl/core/pci.c
> @@ -471,7 +471,7 @@ static int cxl_cdat_get_length(struct device *dev,
> size_t *length)
> {
> __le32 request = CDAT_DOE_REQ(0);
> - __le32 response[32];
> + __le32 response[2];
> int rc;
>
> rc = pci_doe(cdat_doe, PCI_DVSEC_VENDOR_ID_CXL,
> @@ -495,28 +495,28 @@ static int cxl_cdat_read_table(struct device *dev,
> struct pci_doe_mb *cdat_doe,
> struct cxl_cdat *cdat)
> {
> - size_t length = cdat->length;
> - u32 *data = cdat->table;
> + size_t length = cdat->length + sizeof(u32);
> + __le32 *data = cdat->table;
Ah. Makes my earlier comment on this type irrelevant.
> int entry_handle = 0;
> + __le32 saved_dw = 0;
>
> do {
> __le32 request = CDAT_DOE_REQ(entry_handle);
> struct cdat_entry_header *entry;
> - __le32 response[32];
> size_t entry_dw;
> int rc;
>
> rc = pci_doe(cdat_doe, PCI_DVSEC_VENDOR_ID_CXL,
> CXL_DOE_PROTOCOL_TABLE_ACCESS,
> &request, sizeof(request),
> - &response, sizeof(response));
> + data, length);
> if (rc < 0) {
> dev_err(dev, "DOE failed: %d", rc);
> return rc;
> }
>
> /* 1 DW Table Access Response Header + CDAT entry */
> - entry = (struct cdat_entry_header *)(response + 1);
> + entry = (struct cdat_entry_header *)(data + 1);
> if ((entry_handle == 0 &&
> rc != sizeof(u32) + sizeof(struct cdat_header)) ||
> (entry_handle > 0 &&
> @@ -526,21 +526,22 @@ static int cxl_cdat_read_table(struct device *dev,
>
> /* Get the CXL table access header entry handle */
> entry_handle = FIELD_GET(CXL_DOE_TABLE_ACCESS_ENTRY_HANDLE,
> - le32_to_cpu(response[0]));
> + le32_to_cpu(data[0]));
> entry_dw = rc / sizeof(u32);
> /* Skip Header */
> entry_dw -= 1;
> - entry_dw = min(length / sizeof(u32), entry_dw);
> - /* Prevent length < 1 DW from causing a buffer overflow */
> - if (entry_dw) {
> - memcpy(data, entry, entry_dw * sizeof(u32));
> - length -= entry_dw * sizeof(u32);
> - data += entry_dw;
> - }
> + /*
> + * Table Access Response Header overwrote the last DW of
> + * previous entry, so restore that DW
> + */
> + *data = saved_dw;
> + length -= entry_dw * sizeof(u32);
> + data += entry_dw;
> + saved_dw = *data;
> } while (entry_handle != CXL_DOE_TABLE_ACCESS_LAST_ENTRY);
>
> /* Length in CDAT header may exceed concatenation of CDAT entries */
> - cdat->length -= length;
> + cdat->length -= length - sizeof(u32);
>
> return 0;
> }
> @@ -576,7 +577,8 @@ void read_cdat_data(struct cxl_port *port)
> return;
> }
>
> - port->cdat.table = devm_kzalloc(dev, cdat_length, GFP_KERNEL);
> + port->cdat.table = devm_kzalloc(dev, cdat_length + sizeof(u32),
> + GFP_KERNEL);
> if (!port->cdat.table)
> return;
>
> diff --git a/drivers/cxl/cxl.h b/drivers/cxl/cxl.h
> index 1b1cf459ac77..78f5cae5134c 100644
> --- a/drivers/cxl/cxl.h
> +++ b/drivers/cxl/cxl.h
> @@ -494,7 +494,8 @@ struct cxl_pmem_region {
> * @component_reg_phys: component register capability base address (optional)
> * @dead: last ep has been removed, force port re-creation
> * @depth: How deep this port is relative to the root. depth 0 is the root.
> - * @cdat: Cached CDAT data
> + * @cdat: Cached CDAT data (@table is preceded by 4 null bytes, these are not
> + * included in @length)
> * @cdat_available: Should a CDAT attribute be available in sysfs
> */
> struct cxl_port {
> diff --git a/drivers/cxl/port.c b/drivers/cxl/port.c
> index 5453771bf330..0705343ac5ca 100644
> --- a/drivers/cxl/port.c
> +++ b/drivers/cxl/port.c
> @@ -95,7 +95,7 @@ static ssize_t CDAT_read(struct file *filp, struct kobject *kobj,
> return 0;
>
> return memory_read_from_buffer(buf, count, &offset,
> - port->cdat.table,
> + port->cdat.table + sizeof(u32),
> port->cdat.length);
> }
>
next prev parent reply other threads:[~2023-02-14 13:06 UTC|newest]
Thread overview: 62+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-02-10 20:25 [PATCH v3 00/16] Collection of DOE material Lukas Wunner
2023-02-10 20:25 ` [PATCH v3 01/16] cxl/pci: Fix CDAT retrieval on big endian Lukas Wunner
2023-02-11 0:22 ` Dan Williams
2023-02-19 13:03 ` Lukas Wunner
2023-02-14 11:15 ` Jonathan Cameron
2023-02-14 13:51 ` Lukas Wunner
2023-02-14 15:45 ` Jonathan Cameron
2023-02-28 2:53 ` Alexey Kardashevskiy
2023-02-28 8:24 ` Lukas Wunner
2023-02-28 12:08 ` Alexey Kardashevskiy
2023-02-10 20:25 ` [PATCH v3 02/16] cxl/pci: Handle truncated CDAT header Lukas Wunner
2023-02-11 0:40 ` Dan Williams
2023-02-11 9:34 ` Lukas Wunner
2023-02-14 11:16 ` Jonathan Cameron
2023-02-15 1:41 ` Li, Ming
2023-02-10 20:25 ` [PATCH v3 03/16] cxl/pci: Handle truncated CDAT entries Lukas Wunner
2023-02-11 0:50 ` Dan Williams
2023-02-11 10:56 ` Lukas Wunner
2023-02-14 11:30 ` Jonathan Cameron
2023-02-10 20:25 ` [PATCH v3 04/16] cxl/pci: Handle excessive CDAT length Lukas Wunner
2023-02-11 1:04 ` Dan Williams
2023-02-14 11:33 ` Jonathan Cameron
2023-02-16 10:26 ` Lukas Wunner
2023-02-17 10:01 ` Jonathan Cameron
2023-02-10 20:25 ` [PATCH v3 05/16] PCI/DOE: Silence WARN splat with CONFIG_DEBUG_OBJECTS=y Lukas Wunner
2023-02-10 20:25 ` [PATCH v3 06/16] PCI/DOE: Fix memory leak " Lukas Wunner
2023-02-11 1:06 ` Dan Williams
2023-03-01 1:51 ` Davidlohr Bueso
2023-02-10 20:25 ` [PATCH v3 07/16] PCI/DOE: Provide synchronous API and use it internally Lukas Wunner
2023-02-15 1:45 ` Li, Ming
2023-02-28 18:58 ` Davidlohr Bueso
2023-02-10 20:25 ` [PATCH v3 08/16] cxl/pci: Use synchronous API for DOE Lukas Wunner
2023-02-10 20:25 ` [PATCH v3 09/16] PCI/DOE: Make asynchronous API private Lukas Wunner
2023-02-15 1:48 ` Li, Ming
2023-02-10 20:25 ` [PATCH v3 10/16] PCI/DOE: Deduplicate mailbox flushing Lukas Wunner
2023-02-14 11:36 ` Jonathan Cameron
2023-02-15 5:07 ` Li, Ming
2023-02-10 20:25 ` [PATCH v3 11/16] PCI/DOE: Allow mailbox creation without devres management Lukas Wunner
2023-02-14 11:51 ` Jonathan Cameron
2023-02-15 5:17 ` Li, Ming
2023-02-10 20:25 ` [PATCH v3 12/16] PCI/DOE: Create mailboxes on device enumeration Lukas Wunner
2023-02-15 2:07 ` Li, Ming
2023-02-28 1:18 ` Alexey Kardashevskiy
2023-02-28 1:39 ` Dan Williams
2023-02-28 5:43 ` Lukas Wunner
2023-02-28 7:24 ` Alexey Kardashevskiy
2023-02-28 10:42 ` Jonathan Cameron
2023-03-02 20:22 ` Lukas Wunner
2023-03-07 1:55 ` Alexey Kardashevskiy
2023-04-03 0:55 ` Alexey Kardashevskiy
2023-02-10 20:25 ` [PATCH v3 13/16] cxl/pci: Use CDAT DOE mailbox created by PCI core Lukas Wunner
2023-02-10 20:25 ` [PATCH v3 14/16] PCI/DOE: Make mailbox creation API private Lukas Wunner
2023-02-15 2:13 ` Li, Ming
2023-02-10 20:25 ` [PATCH v3 15/16] PCI/DOE: Relax restrictions on request and response size Lukas Wunner
2023-02-15 5:05 ` Li, Ming
2023-02-15 11:49 ` Lukas Wunner
2023-02-10 20:25 ` [PATCH v3 16/16] cxl/pci: Rightsize CDAT response allocation Lukas Wunner
2023-02-14 13:05 ` Jonathan Cameron [this message]
2023-02-16 0:56 ` Ira Weiny
2023-02-16 8:03 ` Lukas Wunner
2023-02-28 1:45 ` Alexey Kardashevskiy
2023-02-28 5:55 ` Lukas Wunner
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=20230214130552.000030ea@Huawei.com \
--to=jonathan.cameron@huawei.com \
--cc=alison.schofield@intel.com \
--cc=bwidawsk@kernel.org \
--cc=dan.j.williams@intel.com \
--cc=dave.jiang@intel.com \
--cc=gregory.price@memverge.com \
--cc=hdanton@sina.com \
--cc=helgaas@kernel.org \
--cc=ira.weiny@intel.com \
--cc=linux-cxl@vger.kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=linuxarm@huawei.com \
--cc=lukas@wunner.de \
--cc=ming4.li@intel.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.