linux-pci.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Niklas Cassel <cassel@kernel.org>
To: "Manivannan Sadhasivam" <manivannan.sadhasivam@linaro.org>,
	"Krzysztof Wilczyński" <kw@linux.com>,
	"Kishon Vijay Abraham I" <kishon@kernel.org>,
	"Arnd Bergmann" <arnd@arndb.de>,
	"Greg Kroah-Hartman" <gregkh@linuxfoundation.org>
Cc: Kuppuswamy Sathyanarayanan
	<sathyanarayanan.kuppuswamy@linux.intel.com>,
	Damien Le Moal <dlemoal@kernel.org>,
	Niklas Cassel <cassel@kernel.org>,
	linux-pci@vger.kernel.org
Subject: [PATCH v4] misc: pci_endpoint_test: Use memcpy_toio()/memcpy_fromio() for BAR tests
Date: Fri, 22 Mar 2024 17:41:38 +0100	[thread overview]
Message-ID: <20240322164139.678228-1-cassel@kernel.org> (raw)

The current code uses writel()/readl(), which has an implicit memory
barrier for every single readl()/writel().

Additionally, reading 4 bytes at a time over the PCI bus is not really
optimal, considering that this code is running in an ioctl handler.

Use memcpy_toio()/memcpy_fromio() for BAR tests.

Before patch with a 4MB BAR:
$ time /usr/bin/pcitest -b 1
BAR1:           OKAY
real    0m 1.56s

After patch with a 4MB BAR:
$ time /usr/bin/pcitest -b 1
BAR1:           OKAY
real    0m 0.54s

Signed-off-by: Niklas Cassel <cassel@kernel.org>
---
Changes since v3:
-Use scope-based resource management __free attribute from cleanup.h to
 avoid overly verbose gotos and labels for error handling.
-Added a comment related to why we allocate a buffer of max 1MB.
 (kmalloc() default upper limit is usually 4 MB on ARM and x86.)

 drivers/misc/pci_endpoint_test.c | 54 +++++++++++++++++++++++++-------
 1 file changed, 42 insertions(+), 12 deletions(-)

diff --git a/drivers/misc/pci_endpoint_test.c b/drivers/misc/pci_endpoint_test.c
index 705029ad8eb5..bf64d3aff7d8 100644
--- a/drivers/misc/pci_endpoint_test.c
+++ b/drivers/misc/pci_endpoint_test.c
@@ -7,6 +7,7 @@
  */
 
 #include <linux/crc32.h>
+#include <linux/cleanup.h>
 #include <linux/delay.h>
 #include <linux/fs.h>
 #include <linux/io.h>
@@ -272,31 +273,60 @@ static const u32 bar_test_pattern[] = {
 	0xA5A5A5A5,
 };
 
+static int pci_endpoint_test_bar_memcmp(struct pci_endpoint_test *test,
+					enum pci_barno barno, int offset,
+					void *write_buf, void *read_buf,
+					int size)
+{
+	memset(write_buf, bar_test_pattern[barno], size);
+	memcpy_toio(test->bar[barno] + offset, write_buf, size);
+
+	memcpy_fromio(read_buf, test->bar[barno] + offset, size);
+
+	return memcmp(write_buf, read_buf, size);
+}
+
 static bool pci_endpoint_test_bar(struct pci_endpoint_test *test,
 				  enum pci_barno barno)
 {
-	int j;
-	u32 val;
-	int size;
+	int j, bar_size, buf_size, iters, remain;
+	void *write_buf __free(kfree) = NULL;
+	void *read_buf __free(kfree) = NULL;
 	struct pci_dev *pdev = test->pdev;
 
 	if (!test->bar[barno])
 		return false;
 
-	size = pci_resource_len(pdev, barno);
+	bar_size = pci_resource_len(pdev, barno);
 
 	if (barno == test->test_reg_bar)
-		size = 0x4;
+		bar_size = 0x4;
 
-	for (j = 0; j < size; j += 4)
-		pci_endpoint_test_bar_writel(test, barno, j,
-					     bar_test_pattern[barno]);
+	/*
+	 * Allocate a buffer of max size 1MB, and reuse that buffer while
+	 * iterating over the whole BAR size (which might be much larger).
+	 */
+	buf_size = min(SZ_1M, bar_size);
 
-	for (j = 0; j < size; j += 4) {
-		val = pci_endpoint_test_bar_readl(test, barno, j);
-		if (val != bar_test_pattern[barno])
+	write_buf = kmalloc(buf_size, GFP_KERNEL);
+	if (!write_buf)
+		return false;
+
+	read_buf = kmalloc(buf_size, GFP_KERNEL);
+	if (!read_buf)
+		return false;
+
+	iters = bar_size / buf_size;
+	for (j = 0; j < iters; j++)
+		if (pci_endpoint_test_bar_memcmp(test, barno, buf_size * j,
+						 write_buf, read_buf, buf_size))
+			return false;
+
+	remain = bar_size % buf_size;
+	if (remain)
+		if (pci_endpoint_test_bar_memcmp(test, barno, buf_size * iters,
+						 write_buf, read_buf, remain))
 			return false;
-	}
 
 	return true;
 }
-- 
2.44.0


             reply	other threads:[~2024-03-22 16:43 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-03-22 16:41 Niklas Cassel [this message]
2024-03-22 17:03 ` [PATCH v4] misc: pci_endpoint_test: Use memcpy_toio()/memcpy_fromio() for BAR tests Kuppuswamy Sathyanarayanan
2024-03-25  7:58   ` Niklas Cassel
2024-04-18  9:08 ` Niklas Cassel
2024-04-18 17:49 ` Kuppuswamy Sathyanarayanan
2024-05-04 14:34 ` Manivannan Sadhasivam

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=20240322164139.678228-1-cassel@kernel.org \
    --to=cassel@kernel.org \
    --cc=arnd@arndb.de \
    --cc=dlemoal@kernel.org \
    --cc=gregkh@linuxfoundation.org \
    --cc=kishon@kernel.org \
    --cc=kw@linux.com \
    --cc=linux-pci@vger.kernel.org \
    --cc=manivannan.sadhasivam@linaro.org \
    --cc=sathyanarayanan.kuppuswamy@linux.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;
as well as URLs for NNTP newsgroup(s).