Skip to contentSkip to conversation

misc: xilinx-sdfec: convert get_user_pages() --> pin_user_pages()

PATCH In review John Hubbard 2 participants 6 messages May 27, 2020 – May 29, 2020

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover misc: xilinx-sdfec: convert get_user_pages() --> pin_user_pages()

Hi,

There are also a couple of tiny cleanup patches, just to fix up a few minor issues that I spotted while converting from get_user_pages_fast() to pin_user_pages_fast().

Note that I have only compile-tested these patches, although that does also include cross-compiling for a few other arches. Any run-time testing would be greatly appreciated!

Cc Derek Kiernan <[email protected]>
Cc Dragan Cvetic <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Michal Simek <[email protected]>

John Hubbard (3):
misc: xilinx-sdfec: improve get_user_pages_fast() error handling
misc: xilinx-sdfec: cleanup return value in xsdfec_table_write()
misc: xilinx-sdfec: convert get_user_pages() --> pin_user_pages()

 drivers/misc/xilinx_sdfec.c | 30 +++++++++++++++++-------------
 1 file changed, 17 insertions(+), 13 deletions(-)

base-commit: 9cb1fd0efd195590b828b9b865421ad345a4a145

Signature · 7 lines
2.26.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#1 lore raw

PATCH 1/3 misc: xilinx-sdfec: improve get_user_pages_fast() error handling

This fixes the case of get_user_pages_fast() returning a -errno. The result needs to be stored in a signed integer. And for safe signed/unsigned comparisons, it's best to keep everything signed. And get_user_pages_fast() also expects a signed value for number of pages to pin.

Therefore, change most relevant variables, from u32 to int. Leave "n" unsigned, for convenience in checking for overflow. And provide a WARN_ON_ONCE() and early return, if overflow occurs.

Also, as long as we're tidying up: rename the page array from page, to pages, in order to match the conventions used in most other call sites.

Fixes 20ec628e8007e ("misc: xilinx_sdfec: Add ability to configure LDPC")
Cc Derek Kiernan <[email protected]>
Cc Dragan Cvetic <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Michal Simek <[email protected]>
---
 drivers/misc/xilinx_sdfec.c | 27 +++++++++++++++++----------
 1 file changed, 17 insertions(+), 10 deletions(-)
drivers/misc/xilinx_sdfec.c +17 -10
index 71bbaa56bdb5..e2766aad9e14 100644
@@ -602,10 +602,10 @@ static int xsdfec_table_write(struct xsdfec_dev *xsdfec, u32 offset,
 			      const u32 depth)
 {
 	u32 reg = 0;
-	u32 res;
-	u32 n, i;
+	int res, i, nr_pages;
+	u32 n;
 	u32 *addr = NULL;
-	struct page *page[MAX_NUM_PAGES];
+	struct page *pages[MAX_NUM_PAGES];
 
 	/*
 	 * Writes that go beyond the length of
@@ -622,15 +622,22 @@ static int xsdfec_table_write(struct xsdfec_dev *xsdfec, u32 offset,
 	if ((len * XSDFEC_REG_WIDTH_JUMP) % PAGE_SIZE)
 		n += 1;
 
-	res = get_user_pages_fast((unsigned long)src_ptr, n, 0, page);
-	if (res < n) {
-		for (i = 0; i < res; i++)
-			put_page(page[i]);
+	if (WARN_ON_ONCE(n > INT_MAX))
+		return -EINVAL;
+
+	nr_pages = n;
+
+	res = get_user_pages_fast((unsigned long)src_ptr, nr_pages, 0, pages);
+	if (res < nr_pages) {
+		if (res > 0) {
+			for (i = 0; i < res; i++)
+				put_page(pages[i]);
+		}
 		return -EINVAL;
 	}
 
-	for (i = 0; i < n; i++) {
-		addr = kmap(page[i]);
+	for (i = 0; i < nr_pages; i++) {
+		addr = kmap(pages[i]);
 		do {
 			xsdfec_regwrite(xsdfec,
 					base_addr + ((offset + reg) *
@@ -639,7 +646,7 @@ static int xsdfec_table_write(struct xsdfec_dev *xsdfec, u32 offset,
 			reg++;
 		} while ((reg < len) &&
 			 ((reg * XSDFEC_REG_WIDTH_JUMP) % PAGE_SIZE));
-		put_page(page[i]);
+		put_page(pages[i]);
 	}
 	return reg;
 }
Signature · 1 line
2.26.2
#2 lore raw

PATCH 2/3 misc: xilinx-sdfec: cleanup return value in xsdfec_table_write()

In reply to John Hubbard

Return 0 for success, rather than the value of an incrementing "reg" index. The reg value was never actually used, so this simplifies the caller slightly.

Cc Derek Kiernan <[email protected]>
Cc Dragan Cvetic <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Michal Simek <[email protected]>
---
 drivers/misc/xilinx_sdfec.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
drivers/misc/xilinx_sdfec.c +1 -3
index e2766aad9e14..7e2ee3e547f2 100644
@@ -648,7 +648,7 @@ static int xsdfec_table_write(struct xsdfec_dev *xsdfec, u32 offset,
 			 ((reg * XSDFEC_REG_WIDTH_JUMP) % PAGE_SIZE));
 		put_page(pages[i]);
 	}
-	return reg;
+	return 0;
 }
 
 static int xsdfec_add_ldpc(struct xsdfec_dev *xsdfec, void __user *arg)
@@ -727,8 +727,6 @@ static int xsdfec_add_ldpc(struct xsdfec_dev *xsdfec, void __user *arg)
 	ret = xsdfec_table_write(xsdfec, 4 * ldpc->qc_off, ldpc->qc_table,
 				 ldpc->nqc, XSDFEC_LDPC_QC_TABLE_ADDR_BASE,
 				 XSDFEC_QC_TABLE_DEPTH);
-	if (ret > 0)
-		ret = 0;
 err_out:
 	kfree(ldpc);
 	return ret;
Signature · 1 line
2.26.2
#3 lore raw

PATCH 3/3 misc: xilinx-sdfec: convert get_user_pages() --> pin_user_pages()

In reply to John Hubbard

This code was using get_user_pages*(), in approximately a "Case 1" scenario (Direct IO), using the categorization from [1]. That means that it's time to convert the get_user_pages*() + put_page() calls to pin_user_pages*() + unpin_user_pages() calls.

There is some helpful background in [2]: basically, this is a small part of fixing a long-standing disconnect between pinning pages, and file systems' use of those pages.

[1] Documentation/core-api/pin_user_pages.rst

[2] "Explicit pinning of user-space pages":

    https://lwn.net/Articles/807108/
Cc Derek Kiernan <[email protected]>
Cc Dragan Cvetic <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Michal Simek <[email protected]>
---
 drivers/misc/xilinx_sdfec.c | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)
drivers/misc/xilinx_sdfec.c +5 -6
index 7e2ee3e547f2..cda3559025d5 100644
@@ -627,12 +627,11 @@ static int xsdfec_table_write(struct xsdfec_dev *xsdfec, u32 offset,
 
 	nr_pages = n;
 
-	res = get_user_pages_fast((unsigned long)src_ptr, nr_pages, 0, pages);
+	res = pin_user_pages_fast((unsigned long)src_ptr, nr_pages, 0, pages);
 	if (res < nr_pages) {
-		if (res > 0) {
-			for (i = 0; i < res; i++)
-				put_page(pages[i]);
-		}
+		if (res > 0)
+			unpin_user_pages(pages, res);
+
 		return -EINVAL;
 	}
 
@@ -646,7 +645,7 @@ static int xsdfec_table_write(struct xsdfec_dev *xsdfec, u32 offset,
 			reg++;
 		} while ((reg < len) &&
 			 ((reg * XSDFEC_REG_WIDTH_JUMP) % PAGE_SIZE));
-		put_page(pages[i]);
+		unpin_user_page(pages[i]);
 	}
 	return 0;
 }
Signature · 1 line
2.26.2
#4 lore raw
In reply to John Hubbard

Hi John,

Thank you for the suggestion, please find my comment below:

Quoted 26 lines

-----Original Message-----
From: John Hubbard <[email protected]>
Sent: Wednesday 27 May 2020 02:26
To: LKML <[email protected]>
Cc: Souptick Joarder <[email protected]>; John Hubbard <[email protected]>; Derek Kiernan <[email protected]>; Dragan Cvetic <[email protected]>; Arnd Bergmann <[email protected]>; Greg Kroah-Hartman <[email protected]>; Michal Simek <[email protected]>; [email protected]
Subject: [PATCH 0/3] misc: xilinx-sdfec: convert get_user_pages() --> pin_user_pages()

Hi,

There are also a couple of tiny cleanup patches, just to fix up a few minor issues that I spotted while converting from get_user_pages_fast() to pin_user_pages_fast().

Note that I have only compile-tested these patches, although that does also include cross-compiling for a few other arches. Any run-time testing would be greatly appreciated!

Cc Derek Kiernan <[email protected]>
Cc Dragan Cvetic <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Michal Simek <[email protected]>
Quoted 4 lines

John Hubbard (3):
misc: xilinx-sdfec: improve get_user_pages_fast() error handling
misc: xilinx-sdfec: cleanup return value in xsdfec_table_write()
misc: xilinx-sdfec: convert get_user_pages() --> pin_user_pages()

Reviewed-by:
Technically there is no problem in this patch, but as you said this should be tested. Currently due to Covid-19 I'm not able to access the HW and I cannot validate this suggestion.

Quoted 8 lines
 drivers/misc/xilinx_sdfec.c | 30 +++++++++++++++++-------------
 1 file changed, 17 insertions(+), 13 deletions(-)

base-commit: 9cb1fd0efd195590b828b9b865421ad345a4a145 --
2.26.2

#5 lore raw
Quoted 36 lines · Dragan Cvetic

Hi John,

Thank you for the suggestion, please find my comment below:

Quoted 26 lines

-----Original Message-----
From: John Hubbard <[email protected]>
Sent: Wednesday 27 May 2020 02:26
To: LKML <[email protected]>
Cc: Souptick Joarder <[email protected]>; John Hubbard <[email protected]>; Derek Kiernan <[email protected]>; Dragan Cvetic <[email protected]>; Arnd Bergmann <[email protected]>; Greg Kroah-Hartman <[email protected]>; Michal Simek <[email protected]>; [email protected]
Subject: [PATCH 0/3] misc: xilinx-sdfec: convert get_user_pages() --> pin_user_pages()

Hi,

There are also a couple of tiny cleanup patches, just to fix up a few minor issues that I spotted while converting from get_user_pages_fast() to pin_user_pages_fast().

Note that I have only compile-tested these patches, although that does also include cross-compiling for a few other arches. Any run-time testing would be greatly appreciated!

Cc Derek Kiernan <[email protected]>
Cc Dragan Cvetic <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Michal Simek <[email protected]>
Quoted 4 lines

John Hubbard (3):
misc: xilinx-sdfec: improve get_user_pages_fast() error handling
misc: xilinx-sdfec: cleanup return value in xsdfec_table_write()
misc: xilinx-sdfec: convert get_user_pages() --> pin_user_pages()

Quoted 4 lines

Reviewed-by:
Technically there is no problem in this patch, but as you said this should be tested. Currently due to Covid-19 I'm not able to access the HW and I cannot validate this suggestion.

Hi Dragan,

Thanks for the review, and for *wanting* to do the testing, even though you can't right now. :)

thanks,

Signature · 2 lines
John Hubbard
NVIDIA
Quoted 9 lines
Quoted 8 lines
  drivers/misc/xilinx_sdfec.c | 30 +++++++++++++++++-------------
  1 file changed, 17 insertions(+), 13 deletions(-)

base-commit: 9cb1fd0efd195590b828b9b865421ad345a4a145 --
2.26.2