[PATCH v2 0/2] USB: replace __get_free_pages() with kmalloc()

[PATCH v2 0/2] USB: replace __get_free_pages() with kmalloc()

[Mike Rapoport (Microsoft)]

This is a (small) part of larger work of replacing page allocator calls
with kmalloc.

My initial intention a few month ago was to remove ugly casts [1], but then
willy pointed out that Linus objected to something like this [2] and it
looks like more than a decade old technical debt.

Largely, anything that doesn't need struct page (or a memdesc in the
future) should just use kmalloc() or kvmalloc() to allocate memory.
kmalloc() guarantees alignment, physical contiguity and working
virt_to_phys() and beside nicer API that returns void * on alloc and
doesn't require to know the allocation size on free, kmalloc() provides
better debugging capabilities than page allocator.

Another thing is that touching these allocation sites gives the reviewers
opportunity to see if a PAGE_SIZE buffer is actually needed or maybe
another size is appropriate.

For larger allocations that don't need physically contiguous memory
kvmalloc() can be a better option that __get_free_pages() because under
memory pressure it's is easier to allocate several order-0 pages than a
physically contiguous chunk with the same number of pages.

And last, but not least, removing needless calls to page allocator should
help with memdesc (aka project folio) conversion. There will be way less
places to audit to see if the user was actually using struct page.

Also in git:
https://git.kernel.org/pub/scm/linux/kernel/git/rppt/linux.git gfp-to-kmalloc/usb

[1] https://lore.kernel.org/all/20251018093002.3660549-1-rppt@kernel.org/
[2] https://lore.kernel.org/all/CA+55aFwp4iy4rtX2gE2WjBGFL=NxMVnoFeHqYa2j1dYOMMGqxg@mail.gmail.com/

---
v2 changes:
- rephrase the sentence about unconditional kfree() to make it clear why
  it is safe
- don't include <linux/slab.h> in ohci-dbg
- fix name of debugfs directory

v1: https://patch.msgid.link/20260630-b4-usb-v1-0-8d547235c374@kernel.org

---
Mike Rapoport (Microsoft) (2):
      usb: host: ohci-dbg: use kmalloc() for print buffer
      usb: core: devices: use kmalloc() to allocate dump buffer

 drivers/usb/core/devices.c  | 7 ++++---
 drivers/usb/host/ohci-dbg.c | 9 +++------
 2 files changed, 7 insertions(+), 9 deletions(-)
---
base-commit: dc59e4fea9d83f03bad6bddf3fa2e52491777482
change-id: 20260616-b4-usb-b302cae4004d

Best regards,
--  
Sincerely yours,
Mike.

[PATCH v2 1/2] usb: host: ohci-dbg: use kmalloc() for print buffer

[Mike Rapoport (Microsoft)]

ochi-dbg allocates buffers for formatting of various dump outputs.

These buffers can be allocated with kmalloc() as there's nothing special
about them to go directly to the page allocator.

kmalloc() provides a better API that does not require ugly casts and
kfree() does not need to know the size of the freed object.

Performance difference between kmalloc() and __get_free_pages() is not
measurable as both allocators take an object/page from a per-CPU list for
fast path allocations.

For the slow path the performance is anyway determined by the amount of
reclaim involved rather than by what allocator is used.

Replace use of get_zeroed_page() with kzalloc() and free_page() with
kfree().

While on it, drop the NULL checks in debug_close(). buf is never NULL
here because all the open handlers return -ENOMEM when alloc_buffer()
fails, and kfree() can handle a NULL buf->page.

Link: https://lore.kernel.org/all/635405e4-9423-4a25-a6e7-e03c8ea0bcbe@redhat.com
Signed-off-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
---
 drivers/usb/host/ohci-dbg.c | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

diff --git a/drivers/usb/host/ohci-dbg.c b/drivers/usb/host/ohci-dbg.c
index 9e0e06bbc570..23dc9eddc06c 100644
--- a/drivers/usb/host/ohci-dbg.c
+++ b/drivers/usb/host/ohci-dbg.c
@@ -683,7 +683,7 @@ static int fill_buffer(struct debug_buffer *buf)
 	int ret;
 
 	if (!buf->page)
-		buf->page = (char *)get_zeroed_page(GFP_KERNEL);
+		buf->page = kzalloc(PAGE_SIZE, GFP_KERNEL);
 
 	if (!buf->page) {
 		ret = -ENOMEM;
@@ -729,11 +729,8 @@ static int debug_close(struct inode *inode, struct file *file)
 {
 	struct debug_buffer *buf = file->private_data;
 
-	if (buf) {
-		if (buf->page)
-			free_page((unsigned long)buf->page);
-		kfree(buf);
-	}
+	kfree(buf->page);
+	kfree(buf);
 
 	return 0;
 }

-- 
2.53.0

[PATCH v2 2/2] usb: core: devices: use kmalloc() to allocate dump buffer

[Mike Rapoport (Microsoft)]

usb_device_dump() allocates a buffer for formatting
/sys/kernel/debug/usb/devices output text.

This buffer can be allocated with kmalloc() as there's nothing special
about it to go directly to the page allocator.

kmalloc() provides a better API that does not require ugly casts and
kfree() does not need to know the size of the freed object.

Performance difference between kmalloc() and __get_free_pages() is not
measurable as both allocators take an object/page from a per-CPU list for
fast path allocations.

For the slow path the performance is anyway determined by the amount of
reclaim involved rather than by what allocator is used.

Replace use of __get_free_pages() with kmalloc() and free_pages() with
kfree().

Link: https://lore.kernel.org/all/635405e4-9423-4a25-a6e7-e03c8ea0bcbe@redhat.com
Signed-off-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
Reviewed-by: Alan Stern <stern@rowland.harvard.edu>
---
 drivers/usb/core/devices.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/usb/core/devices.c b/drivers/usb/core/devices.c
index a247da73f34d..6f0354aba38b 100644
--- a/drivers/usb/core/devices.c
+++ b/drivers/usb/core/devices.c
@@ -37,6 +37,7 @@
  */
 
 #include <linux/fs.h>
+#include <linux/slab.h>
 #include <linux/mm.h>
 #include <linux/gfp.h>
 #include <linux/usb.h>
@@ -408,7 +409,7 @@ static ssize_t usb_device_dump(char __user **buffer, size_t *nbytes,
 		return 0;
 	/* allocate 2^1 pages = 8K (on i386);
 	 * should be more than enough for one device */
-	pages_start = (char *)__get_free_pages(GFP_NOIO, 1);
+	pages_start = kmalloc(PAGE_SIZE << 1, GFP_NOIO);
 	if (!pages_start)
 		return -ENOMEM;
 
@@ -479,7 +480,7 @@ static ssize_t usb_device_dump(char __user **buffer, size_t *nbytes,
 		if (length > *nbytes)
 			length = *nbytes;
 		if (copy_to_user(*buffer, pages_start + *skip_bytes, length)) {
-			free_pages((unsigned long)pages_start, 1);
+			kfree(pages_start);
 			return -EFAULT;
 		}
 		*nbytes -= length;
@@ -490,7 +491,7 @@ static ssize_t usb_device_dump(char __user **buffer, size_t *nbytes,
 	} else
 		*skip_bytes -= length;
 
-	free_pages((unsigned long)pages_start, 1);
+	kfree(pages_start);
 
 	/* Now look at all of this device's children. */
 	usb_hub_for_each_child(usbdev, chix, childdev) {

-- 
2.53.0

Re: [PATCH v2 1/2] usb: host: ohci-dbg: use kmalloc() for print buffer

[Alan Stern]

On Wed, Jul 01, 2026 at 12:41:50PM +0300, Mike Rapoport (Microsoft) wrote:
> ochi-dbg allocates buffers for formatting of various dump outputs.
> 
> These buffers can be allocated with kmalloc() as there's nothing special
> about them to go directly to the page allocator.
> 
> kmalloc() provides a better API that does not require ugly casts and
> kfree() does not need to know the size of the freed object.
> 
> Performance difference between kmalloc() and __get_free_pages() is not
> measurable as both allocators take an object/page from a per-CPU list for
> fast path allocations.
> 
> For the slow path the performance is anyway determined by the amount of
> reclaim involved rather than by what allocator is used.
> 
> Replace use of get_zeroed_page() with kzalloc() and free_page() with
> kfree().
> 
> While on it, drop the NULL checks in debug_close(). buf is never NULL
> here because all the open handlers return -ENOMEM when alloc_buffer()
> fails, and kfree() can handle a NULL buf->page.
> 
> Link: https://lore.kernel.org/all/635405e4-9423-4a25-a6e7-e03c8ea0bcbe@redhat.com
> Signed-off-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
> ---

Reviewed-by: Alan Stern <stern@rowland.harvard.edu>

Incidentally, I filtered it out of my reply this time, but on this and 
previous submissions you have a CC: entry that says <rpppt@kernel.org>, 
which is an invalid address (three 'p's instead of two).  Is there a 
typo in one of your scripts or settings?  It's a little surprising that 
you aren't constantly getting email error messages every time you send 
something out.

Alan Stern

>  drivers/usb/host/ohci-dbg.c | 9 +++------
>  1 file changed, 3 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/usb/host/ohci-dbg.c b/drivers/usb/host/ohci-dbg.c
> index 9e0e06bbc570..23dc9eddc06c 100644
> --- a/drivers/usb/host/ohci-dbg.c
> +++ b/drivers/usb/host/ohci-dbg.c
> @@ -683,7 +683,7 @@ static int fill_buffer(struct debug_buffer *buf)
>  	int ret;
>  
>  	if (!buf->page)
> -		buf->page = (char *)get_zeroed_page(GFP_KERNEL);
> +		buf->page = kzalloc(PAGE_SIZE, GFP_KERNEL);
>  
>  	if (!buf->page) {
>  		ret = -ENOMEM;
> @@ -729,11 +729,8 @@ static int debug_close(struct inode *inode, struct file *file)
>  {
>  	struct debug_buffer *buf = file->private_data;
>  
> -	if (buf) {
> -		if (buf->page)
> -			free_page((unsigned long)buf->page);
> -		kfree(buf);
> -	}
> +	kfree(buf->page);
> +	kfree(buf);
>  
>  	return 0;
>  }
> 
> -- 
> 2.53.0
> 

Re: [PATCH v2 1/2] usb: host: ohci-dbg: use kmalloc() for print buffer

[Mike Rapoport]

On Wed, Jul 01, 2026 at 09:52:42AM -0400, Alan Stern wrote:
> On Wed, Jul 01, 2026 at 12:41:50PM +0300, Mike Rapoport (Microsoft) wrote:
> > ochi-dbg allocates buffers for formatting of various dump outputs.
> > 
> > These buffers can be allocated with kmalloc() as there's nothing special
> > about them to go directly to the page allocator.
> > 
> > kmalloc() provides a better API that does not require ugly casts and
> > kfree() does not need to know the size of the freed object.
> > 
> > Performance difference between kmalloc() and __get_free_pages() is not
> > measurable as both allocators take an object/page from a per-CPU list for
> > fast path allocations.
> > 
> > For the slow path the performance is anyway determined by the amount of
> > reclaim involved rather than by what allocator is used.
> > 
> > Replace use of get_zeroed_page() with kzalloc() and free_page() with
> > kfree().
> > 
> > While on it, drop the NULL checks in debug_close(). buf is never NULL
> > here because all the open handlers return -ENOMEM when alloc_buffer()
> > fails, and kfree() can handle a NULL buf->page.
> > 
> > Link: https://lore.kernel.org/all/635405e4-9423-4a25-a6e7-e03c8ea0bcbe@redhat.com
> > Signed-off-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
> > ---
> 
> Reviewed-by: Alan Stern <stern@rowland.harvard.edu>

Thanks!
 
> Incidentally, I filtered it out of my reply this time, but on this and 
> previous submissions you have a CC: entry that says <rpppt@kernel.org>, 
> which is an invalid address (three 'p's instead of two).  Is there a 
> typo in one of your scripts or settings?  It's a little surprising that 
> you aren't constantly getting email error messages every time you send 
> something out.

I'm gradually switching to b4 send and this looks like off-by-one in my
manual editing of the addresses.
 
> Alan Stern

-- 
Sincerely yours,
Mike.