Skip to contentSkip to conversation

USB: add __get_free_pages() in hcd_buffer_alloc function

PATCH In review [email protected] 3 participants 5 messages Jan 3, 2021 USB

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 USB: add __get_free_pages() in hcd_buffer_alloc function

From: Zqiang <[email protected]>

When USB hosts just use PIO, allocate memory by slab/slub memory manager, if the required memory size is larger than one or more page sizes, need allocate memory directly from buddy systems.

---
 drivers/usb/core/buffer.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/usb/core/buffer.c +4 -3
index fbb087b728dc..28e7db9f93d5 100644
@@ -131,7 +131,8 @@ void *hcd_buffer_alloc(
 	/* some USB hosts just use PIO */
 	if (!hcd_uses_dma(hcd)) {
 		*dma = ~(dma_addr_t) 0;
-		return kmalloc(size, mem_flags);
+		return size < PAGE_SIZE ? kmalloc(size, mem_flags)
+					: __get_free_pages(mem_flags, get_order(size));
 	}
 
 	for (i = 0; i < HCD_BUFFER_POOLS; i++) {
@@ -160,8 +161,8 @@ void hcd_buffer_free(
 	}
 
 	if (!hcd_uses_dma(hcd)) {
-		kfree(addr);
-		return;
+		return size < PAGE_SIZE ? kfree(addr)
+					: free_pages(addr, get_order(size));
 	}
 
 	for (i = 0; i < HCD_BUFFER_POOLS; i++) {
Signature · 1 line
2.17.1
#1 lore raw
Thank you for the patch! Perhaps something to improve:

Hi,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on usb/usb-testing]
[also build test WARNING on v5.11-rc1 next-20201223]
[If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/qiang-zhang-windriver-com/USB-add-__get_free_pages-in-hcd_buffer_alloc-function/20210103-162325
base:   https://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb.git usb-testing
config: x86_64-randconfig-a006-20210103 (attached as .config)
compiler: clang version 12.0.0 (https://github.com/llvm/llvm-project 7af6a134508cd1c7f75c6e3441ce436f220f30a4)

reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# install x86_64 cross compiling tool for clang build
# apt-get install binutils-x86-64-linux-gnu
# https://github.com/0day-ci/linux/commit/904936636c0b20b1d9b2be051502c6d80c5fa1ee
git remote add linux-review https://github.com/0day-ci/linux
git fetch --no-tags linux-review qiang-zhang-windriver-com/USB-add-__get_free_pages-in-hcd_buffer_alloc-function/20210103-162325 git checkout 904936636c0b20b1d9b2be051502c6d80c5fa1ee
# save the attached .config to linux build tree
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross ARCH=x86_64

If you fix the issue, kindly add following tag as appropriate

All warnings (new ones prefixed by >>):

>> drivers/usb/core/buffer.c:134:27: warning: pointer/integer type mismatch in conditional expression ('void *' and 'unsigned long') [-Wconditional-type-mismatch]
                   return size < PAGE_SIZE ? kmalloc(size, mem_flags)
                                           ^ ~~~~~~~~~~~~~~~~~~~~~~~~
>> drivers/usb/core/buffer.c:165:19: warning: incompatible pointer to integer conversion passing 'void *' to parameter of type 'unsigned long' [-Wint-conversion]
                                           : free_pages(addr, get_order(size));
                                                        ^~~~
   include/linux/gfp.h:582:38: note: passing argument to parameter 'addr' here
   extern void free_pages(unsigned long addr, unsigned int order);
                                        ^

2 warnings generated.

vim +134 drivers/usb/core/buffer.c

109
110
111 /* sometimes alloc/free could use kmalloc with GFP_DMA, for
112 * better sharing and to leverage mm/slab.c intelligence.
113 */
114
115 void *hcd_buffer_alloc(
116 struct usb_bus *bus,
117 size_t size,
118 gfp_t mem_flags,
119 dma_addr_t *dma
120 )
121 {
122 struct usb_hcd *hcd = bus_to_hcd(bus);
123 int i;
124
125 if (size == 0)
126 return NULL;
127
128 if (hcd->localmem_pool)
129 return gen_pool_dma_alloc(hcd->localmem_pool, size, dma);
130
131 /* some USB hosts just use PIO */
132 if (!hcd_uses_dma(hcd)) {
133 *dma = ~(dma_addr_t) 0;

 > 134			return size < PAGE_SIZE ? kmalloc(size, mem_flags)
   135						: __get_free_pages(mem_flags, get_order(size));
   136		}
   137	
   138		for (i = 0; i < HCD_BUFFER_POOLS; i++) {
   139			if (size <= pool_max[i])
   140				return dma_pool_alloc(hcd->pool[i], mem_flags, dma);
   141		}
   142		return dma_alloc_coherent(hcd->self.sysdev, size, dma, mem_flags);
   143	}
   144	
   145	void hcd_buffer_free(
   146		struct usb_bus		*bus,
   147		size_t			size,
   148		void			*addr,
   149		dma_addr_t		dma
   150	)
   151	{
   152		struct usb_hcd		*hcd = bus_to_hcd(bus);
   153		int			i;
   154	
   155		if (!addr)
   156			return;
   157	
   158		if (hcd->localmem_pool) {
   159			gen_pool_free(hcd->localmem_pool, (unsigned long)addr, size);
   160			return;
   161		}
   162	
   163		if (!hcd_uses_dma(hcd)) {
   164			return size < PAGE_SIZE ? kfree(addr)
 > 165						: free_pages(addr, get_order(size));

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org

#2 lore raw
In reply to [email protected]
Thank you for the patch! Perhaps something to improve:

Hi,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on usb/usb-testing]
[also build test WARNING on v5.11-rc1 next-20201223]
[If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/qiang-zhang-windriver-com/USB-add-__get_free_pages-in-hcd_buffer_alloc-function/20210103-162325
base:   https://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb.git usb-testing
config: parisc-randconfig-r033-20210103 (attached as .config)
compiler: hppa-linux-gcc (GCC) 9.3.0

reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# https://github.com/0day-ci/linux/commit/904936636c0b20b1d9b2be051502c6d80c5fa1ee
git remote add linux-review https://github.com/0day-ci/linux
git fetch --no-tags linux-review qiang-zhang-windriver-com/USB-add-__get_free_pages-in-hcd_buffer_alloc-function/20210103-162325 git checkout 904936636c0b20b1d9b2be051502c6d80c5fa1ee
# save the attached .config to linux build tree
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-9.3.0 make.cross ARCH=parisc

If you fix the issue, kindly add following tag as appropriate

All warnings (new ones prefixed by >>):

drivers/usb/core/buffer.c: In function 'hcd_buffer_alloc':

>> drivers/usb/core/buffer.c:135:6: warning: pointer/integer type mismatch in conditional expression
     135 |      : __get_free_pages(mem_flags, get_order(size));
         |      ^
   drivers/usb/core/buffer.c: In function 'hcd_buffer_free':
>> drivers/usb/core/buffer.c:165:19: warning: passing argument 1 of 'free_pages' makes integer from pointer without a cast [-Wint-conversion]
     165 |      : free_pages(addr, get_order(size));
         |                   ^~~~
         |                   |
         |                   void *
   In file included from include/linux/umh.h:4,
                    from include/linux/kmod.h:9,
                    from include/linux/module.h:16,
                    from drivers/usb/core/buffer.c:11:
   include/linux/gfp.h:582:38: note: expected 'long unsigned int' but argument is of type 'void *'
     582 | extern void free_pages(unsigned long addr, unsigned int order);
         |                        ~~~~~~~~~~~~~~^~~~

vim +135 drivers/usb/core/buffer.c

109
110
111 /* sometimes alloc/free could use kmalloc with GFP_DMA, for
112 * better sharing and to leverage mm/slab.c intelligence.
113 */
114
115 void *hcd_buffer_alloc(
116 struct usb_bus *bus,
117 size_t size,
118 gfp_t mem_flags,
119 dma_addr_t *dma
120 )
121 {
122 struct usb_hcd *hcd = bus_to_hcd(bus);
123 int i;
124
125 if (size == 0)
126 return NULL;
127
128 if (hcd->localmem_pool)
129 return gen_pool_dma_alloc(hcd->localmem_pool, size, dma);
130
131 /* some USB hosts just use PIO */
132 if (!hcd_uses_dma(hcd)) {
133 *dma = ~(dma_addr_t) 0;
134 return size < PAGE_SIZE ? kmalloc(size, mem_flags)

 > 135						: __get_free_pages(mem_flags, get_order(size));
   136		}
   137	
   138		for (i = 0; i < HCD_BUFFER_POOLS; i++) {
   139			if (size <= pool_max[i])
   140				return dma_pool_alloc(hcd->pool[i], mem_flags, dma);
   141		}
   142		return dma_alloc_coherent(hcd->self.sysdev, size, dma, mem_flags);
   143	}
   144	
   145	void hcd_buffer_free(
   146		struct usb_bus		*bus,
   147		size_t			size,
   148		void			*addr,
   149		dma_addr_t		dma
   150	)
   151	{
   152		struct usb_hcd		*hcd = bus_to_hcd(bus);
   153		int			i;
   154	
   155		if (!addr)
   156			return;
   157	
   158		if (hcd->localmem_pool) {
   159			gen_pool_free(hcd->localmem_pool, (unsigned long)addr, size);
   160			return;
   161		}
   162	
   163		if (!hcd_uses_dma(hcd)) {
   164			return size < PAGE_SIZE ? kfree(addr)
 > 165						: free_pages(addr, get_order(size));

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org

maintainer
#3 lore raw
In reply to [email protected]
Quoted 6 lines · On Sun, Jan 03, 2021 at 04:19:02PM +0800, [email protected] wrote:

From: Zqiang <[email protected]>

When USB hosts just use PIO, allocate memory by slab/slub memory manager, if the required memory size is larger than one or more page sizes, need allocate memory directly from buddy systems.

That says _what_ you are doing, but not _why_ you need to do this.

What in-tree host controller has this problem and on what platform?

Quoted 11 lines
---
 drivers/usb/core/buffer.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/usb/core/buffer.c +0 -0
index fbb087b728dc..28e7db9f93d5 100644

drivers/usb/core/buffer.c void *hcd_buffer_alloc( · line 135

Quoted 6 lines
drivers/usb/core/buffer.c +2 -1
@@ -131,7 +131,8 @@ void *hcd_buffer_alloc(
 	/* some USB hosts just use PIO */
 	if (!hcd_uses_dma(hcd)) {
 		*dma = ~(dma_addr_t) 0;
-		return kmalloc(size, mem_flags);
+		return size < PAGE_SIZE ? kmalloc(size, mem_flags)
+					: __get_free_pages(mem_flags, get_order(size));

Please never use ? : statements if at all possible. Use a real if () instead.

Oh, and make sure your code actually builds properly :(

What platforms did you test this on?

thanks,

greg k-h

#4 lore raw

________________________________________
发件人: Greg KH <[email protected]>
发送时间: 2021年1月3日 19:03
收件人: Zhang, Qiang
抄送: [email protected]; [email protected]
主题: Re: [PATCH] USB: add __get_free_pages() in hcd_buffer_alloc function

Quoted 32 lines · On Sun, Jan 03, 2021 at 04:19:02PM +0800, [email protected] wrote:

From: Zqiang <[email protected]>

When USB hosts just use PIO, allocate memory by slab/slub memory manager, if the required memory size is larger than one or more page sizes, need allocate memory directly from buddy systems.

That says _what_ you are doing, but not _why_ you need to do this.

What in-tree host controller has this problem and on what platform?

---
 drivers/usb/core/buffer.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/usb/core/buffer.c +2 -1
index fbb087b728dc..28e7db9f93d5 100644
@@ -131,7 +131,8 @@ void *hcd_buffer_alloc(
      /* some USB hosts just use PIO */
      if (!hcd_uses_dma(hcd)) {
              *dma = ~(dma_addr_t) 0;
-             return kmalloc(size, mem_flags);
+             return size < PAGE_SIZE ? kmalloc(size, mem_flags)
+                                     : __get_free_pages(mem_flags, get_order(size));
 
 Please never use ? : statements if at all possible.  Use a real if ()

instead.

Quoted 1 line

Oh, and make sure your code actually builds properly :(

I'm very sorry. It was my mistake, I will modify and resend. if some hcd hosts not support DMA and localmem_pool, when we call usb_alloc_coherent function ,It's actually just a call kmalloc function, it is only suitable for allocating memory smaller than one pagesize, if size is larger than the size of a pagesize, need allocate memory directly from buddy system.

Quoted 5 lines

What platforms did you test this on?

thanks,

greg k-h