[PATCH 0/3] ALSA: use vmalloc_array() to simplify code

[PATCH 0/3] ALSA: use vmalloc_array() to simplify code

[Qianfeng Rong]

Remove array_size() calls and replace vmalloc() with vmalloc_array() to
simplify the code and maintain consistency with existing kmalloc_array()
usage.

vmalloc_array() is also optimized better, resulting in less instructions
being used [1].

Link: https://lore.kernel.org/lkml/abc66ec5-85a4-47e1-9759-2f60ab111971@vivo.com/ #1
Signed-off-by: Qianfeng Rong <rongqianfeng@vivo.com>

Qianfeng Rong (3):
  ALSA: cs46xx: use vmalloc_array() to simplify code
  ALSA: emu10k1: use vmalloc_array() to simplify code
  ALSA: maestro3: use vmalloc_array() to simplify code

 sound/pci/cs46xx/dsp_spos.c      | 4 ++--
 sound/pci/emu10k1/emu10k1_main.c | 8 ++++----
 sound/pci/emu10k1/emufx.c        | 2 +-
 sound/pci/emu10k1/p16v.c         | 2 +-
 sound/pci/maestro3.c             | 5 ++---
 5 files changed, 10 insertions(+), 11 deletions(-)

-- 
2.34.1

[PATCH 1/3] ALSA: cs46xx: use vmalloc_array() to simplify code

[Qianfeng Rong]

Remove array_size() calls and replace vmalloc() with vmalloc_array() in
cs46xx_dsp_spos_create() to simplify the code and maintain consistency
with existing kmalloc_array() usage.

Signed-off-by: Qianfeng Rong <rongqianfeng@vivo.com>
---
 sound/pci/cs46xx/dsp_spos.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/sound/pci/cs46xx/dsp_spos.c b/sound/pci/cs46xx/dsp_spos.c
index e07f85322f1c..0aa68f9dd2aa 100644
--- a/sound/pci/cs46xx/dsp_spos.c
+++ b/sound/pci/cs46xx/dsp_spos.c
@@ -228,8 +228,8 @@ struct dsp_spos_instance *cs46xx_dsp_spos_create (struct snd_cs46xx * chip)
 
 	/* better to use vmalloc for this big table */
 	ins->symbol_table.symbols =
-		vmalloc(array_size(DSP_MAX_SYMBOLS,
-				   sizeof(struct dsp_symbol_entry)));
+		vmalloc_array(DSP_MAX_SYMBOLS,
+			      sizeof(struct dsp_symbol_entry));
 	ins->code.data = kmalloc(DSP_CODE_BYTE_SIZE, GFP_KERNEL);
 	ins->modules = kmalloc_array(DSP_MAX_MODULES,
 				     sizeof(struct dsp_module_desc),
-- 
2.34.1

[PATCH 2/3] ALSA: emu10k1: use vmalloc_array() to simplify code

[Qianfeng Rong]

Remove array_size() calls and replace vmalloc() with vmalloc_array() to
simplify the code and maintain consistency with existing kmalloc_array()
usage.

Signed-off-by: Qianfeng Rong <rongqianfeng@vivo.com>
---
 sound/pci/emu10k1/emu10k1_main.c | 8 ++++----
 sound/pci/emu10k1/emufx.c        | 2 +-
 sound/pci/emu10k1/p16v.c         | 2 +-
 3 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/sound/pci/emu10k1/emu10k1_main.c b/sound/pci/emu10k1/emu10k1_main.c
index bbe252b8916c..f6ae78956eca 100644
--- a/sound/pci/emu10k1/emu10k1_main.c
+++ b/sound/pci/emu10k1/emu10k1_main.c
@@ -1579,10 +1579,10 @@ int snd_emu10k1_create(struct snd_card *card,
 		(unsigned long)emu->ptb_pages.addr,
 		(unsigned long)(emu->ptb_pages.addr + emu->ptb_pages.bytes));
 
-	emu->page_ptr_table = vmalloc(array_size(sizeof(void *),
-						 emu->max_cache_pages));
-	emu->page_addr_table = vmalloc(array_size(sizeof(unsigned long),
-						  emu->max_cache_pages));
+	emu->page_ptr_table = vmalloc_array(emu->max_cache_pages,
+					    sizeof(void *));
+	emu->page_addr_table = vmalloc_array(emu->max_cache_pages,
+					     sizeof(unsigned long));
 	if (!emu->page_ptr_table || !emu->page_addr_table)
 		return -ENOMEM;
 
diff --git a/sound/pci/emu10k1/emufx.c b/sound/pci/emu10k1/emufx.c
index 7db0660e6b61..130e713fe248 100644
--- a/sound/pci/emu10k1/emufx.c
+++ b/sound/pci/emu10k1/emufx.c
@@ -2648,7 +2648,7 @@ int snd_emu10k1_efx_alloc_pm_buffer(struct snd_emu10k1 *emu)
 	if (! emu->tram_val_saved || ! emu->tram_addr_saved)
 		return -ENOMEM;
 	len = emu->audigy ? 2 * 1024 : 2 * 512;
-	emu->saved_icode = vmalloc(array_size(len, 4));
+	emu->saved_icode = vmalloc_array(len, 4);
 	if (! emu->saved_icode)
 		return -ENOMEM;
 	return 0;
diff --git a/sound/pci/emu10k1/p16v.c b/sound/pci/emu10k1/p16v.c
index e774174d10de..ca732e6464ec 100644
--- a/sound/pci/emu10k1/p16v.c
+++ b/sound/pci/emu10k1/p16v.c
@@ -795,7 +795,7 @@ int snd_p16v_mixer(struct snd_emu10k1 *emu)
 
 int snd_p16v_alloc_pm_buffer(struct snd_emu10k1 *emu)
 {
-	emu->p16v_saved = vmalloc(array_size(NUM_CHS * 4, 0x80));
+	emu->p16v_saved = vmalloc_array(NUM_CHS * 4, 0x80);
 	if (! emu->p16v_saved)
 		return -ENOMEM;
 	return 0;
-- 
2.34.1

[PATCH 3/3] ALSA: maestro3: use vmalloc_array() to simplify code

[Qianfeng Rong]

Remove array_size() calls and replace vmalloc() with vmalloc_array() in
snd_m3_create() to simplify the code.

Signed-off-by: Qianfeng Rong <rongqianfeng@vivo.com>
---
 sound/pci/maestro3.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/sound/pci/maestro3.c b/sound/pci/maestro3.c
index e092097599ff..0170b432402b 100644
--- a/sound/pci/maestro3.c
+++ b/sound/pci/maestro3.c
@@ -2580,9 +2580,8 @@ snd_m3_create(struct snd_card *card, struct pci_dev *pci,
 
 	if (IS_ENABLED(CONFIG_PM_SLEEP)) {
 		chip->suspend_mem =
-			vmalloc(array_size(sizeof(u16),
-					   REV_B_CODE_MEMORY_LENGTH +
-					   REV_B_DATA_MEMORY_LENGTH));
+			vmalloc_array(REV_B_CODE_MEMORY_LENGTH +
+				      REV_B_DATA_MEMORY_LENGTH, sizeof(u16));
 		if (!chip->suspend_mem)
 			dev_warn(card->dev, "can't allocate apm buffer\n");
 	}
-- 
2.34.1

Re: [PATCH 0/3] ALSA: use vmalloc_array() to simplify code

[Andy Shevchenko]

On Sun, Aug 24, 2025 at 09:55:09PM +0800, Qianfeng Rong wrote:
> Remove array_size() calls and replace vmalloc() with vmalloc_array() to
> simplify the code and maintain consistency with existing kmalloc_array()
> usage.
> 
> vmalloc_array() is also optimized better, resulting in less instructions
> being used [1].

Have you considered using sizeof(*...) where it makes sense?

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH 0/3] ALSA: use vmalloc_array() to simplify code

[Qianfeng Rong]

在 2025/8/25 19:01, Andy Shevchenko 写道:
> On Sun, Aug 24, 2025 at 09:55:09PM +0800, Qianfeng Rong wrote:
>> Remove array_size() calls and replace vmalloc() with vmalloc_array() to
>> simplify the code and maintain consistency with existing kmalloc_array()
>> usage.
>>
>> vmalloc_array() is also optimized better, resulting in less instructions
>> being used [1].
> Have you considered using sizeof(*...) where it makes sense?
I believe that sizeof(*...) should preferably not be mixed with this patch;
instead, it should be addressed in a separate, independent patch that
covers all relevant instances.

Additionally, I understand that not everyone prefers using sizeof(*...), as
it reduces readability.  What do you think, Andy?

Best regards,
Qianfeng
>

Re: [PATCH 0/3] ALSA: use vmalloc_array() to simplify code

[Philipp Stanner]

On Mon, 2025-08-25 at 19:46 +0800, Qianfeng Rong wrote:
> 
> 在 2025/8/25 19:01, Andy Shevchenko 写道:
> > On Sun, Aug 24, 2025 at 09:55:09PM +0800, Qianfeng Rong wrote:
> > > Remove array_size() calls and replace vmalloc() with vmalloc_array() to
> > > simplify the code and maintain consistency with existing kmalloc_array()
> > > usage.
> > > 
> > > vmalloc_array() is also optimized better, resulting in less instructions
> > > being used [1].
> > Have you considered using sizeof(*...) where it makes sense?
> I believe that sizeof(*...) should preferably not be mixed with this patch;
> instead, it should be addressed in a separate, independent patch that
> covers all relevant instances.
> 
> Additionally, I understand that not everyone prefers using sizeof(*...), as
> it reduces readability.  What do you think, Andy?

I agree that should be addressed in a separate patch, if at all.

As for sizeof(*foo), one gets used to it. I think it's preferred in the
kernel often because it's more difficult to introduce bugs like so:

long *foo;

foo = kmalloc(sizeof(long) * 9001);

When you later change long *foo to int *foo and forget about the malloc
place, you'll get an overflow.

foo = kmalloc(sizeof(*foo) …) can never have that problem.

Regards
P.


> 
> Best regards,
> Qianfeng
> >