From: Li RongQing <lirongqing@baidu.com>

The DMAPOOL_DEBUG macro is only defined when CONFIG_SLUB_DEBUG_ON is
enabled, and it is used exclusively to guard the pool_check_block()
function. Wrapping CONFIG_SLUB_DEBUG_ON in another local macro adds
unnecessary indirection.

Simplify the code by removing DMAPOOL_DEBUG entirely and using the
standard CONFIG_SLUB_DEBUG_ON config option directly.

Signed-off-by: Li RongQing <lirongqing@baidu.com>
---
 mm/dmapool.c | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/mm/dmapool.c b/mm/dmapool.c
index 5d8af6e..af43f1a 100644
--- a/mm/dmapool.c
+++ b/mm/dmapool.c
@@ -36,10 +36,6 @@
 #include <linux/types.h>
 #include <linux/wait.h>
 
-#ifdef CONFIG_SLUB_DEBUG_ON
-#define DMAPOOL_DEBUG 1
-#endif
-
 struct dma_block {
 	struct dma_block *next_block;
 	dma_addr_t dma;
@@ -92,7 +88,7 @@ static ssize_t pools_show(struct device *dev, struct device_attribute *attr, cha
 
 static DEVICE_ATTR_RO(pools);
 
-#ifdef DMAPOOL_DEBUG
+#ifdef CONFIG_SLUB_DEBUG_ON
 static void pool_check_block(struct dma_pool *pool, struct dma_block *block,
 			     gfp_t mem_flags)
 {
-- 
2.9.4

On 5/22/26 12:53, lirongqing wrote:
> From: Li RongQing <lirongqing@baidu.com>
> 
> The DMAPOOL_DEBUG macro is only defined when CONFIG_SLUB_DEBUG_ON is
> enabled, and it is used exclusively to guard the pool_check_block()
> function. Wrapping CONFIG_SLUB_DEBUG_ON in another local macro adds
> unnecessary indirection.
> 
> Simplify the code by removing DMAPOOL_DEBUG entirely and using the
> standard CONFIG_SLUB_DEBUG_ON config option directly.
> 
> Signed-off-by: Li RongQing <lirongqing@baidu.com>

I think a more useful change would be to remove the usage of
CONFIG_SLUB_DEBUG_ON completely and rather (re)use a boot option that uses a
static key. CONFIG_SLUB_DEBUG_ON is only useful for bots running
randconfigs, otherwise normally for debugging a specific problem one doesn't
want to rebuild a kernel with CONFIG_SLUB_DEBUG_ON, but just reboot with a
boot option slab_debug[=...] (for slab itself) or debug_pagealloc=on etc.

(same thing should be done for the usage in mm/mempool.c btw)

> ---
>  mm/dmapool.c | 6 +-----
>  1 file changed, 1 insertion(+), 5 deletions(-)
> 
> diff --git a/mm/dmapool.c b/mm/dmapool.c
> index 5d8af6e..af43f1a 100644
> --- a/mm/dmapool.c
> +++ b/mm/dmapool.c
> @@ -36,10 +36,6 @@
>  #include <linux/types.h>
>  #include <linux/wait.h>
>  
> -#ifdef CONFIG_SLUB_DEBUG_ON
> -#define DMAPOOL_DEBUG 1
> -#endif
> -
>  struct dma_block {
>  	struct dma_block *next_block;
>  	dma_addr_t dma;
> @@ -92,7 +88,7 @@ static ssize_t pools_show(struct device *dev, struct device_attribute *attr, cha
>  
>  static DEVICE_ATTR_RO(pools);
>  
> -#ifdef DMAPOOL_DEBUG
> +#ifdef CONFIG_SLUB_DEBUG_ON
>  static void pool_check_block(struct dma_pool *pool, struct dma_block *block,
>  			     gfp_t mem_flags)
>  {

> 
> I think a more useful change would be to remove the usage of
> CONFIG_SLUB_DEBUG_ON completely and rather (re)use a boot option that
> uses a static key. CONFIG_SLUB_DEBUG_ON is only useful for bots running
> randconfigs, otherwise normally for debugging a specific problem one
> doesn't want to rebuild a kernel with CONFIG_SLUB_DEBUG_ON, but just
> reboot with a boot option slab_debug[=...] (for slab itself) or
> debug_pagealloc=on etc.
> 
> (same thing should be done for the usage in mm/mempool.c btw)

Good idea, I will send v2

Thanks
 

[Li,Rongqing]