1. Patchew
  2. linux
  3. View series

[PATCH] RDMA/irdma: Use kvzalloc for paged memory DMA address array

Carlos Bilbao posted 1 patch 2 weeks, 3 days ago
There is a newer version of this series
drivers/infiniband/hw/irdma/utils.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
[PATCH] RDMA/irdma: Use kvzalloc for paged memory DMA address array
Posted by Carlos Bilbao 2 weeks, 3 days ago 
Allocate array chunk->dmainfo.dmaaddrs using kvzalloc() to allow the
allocation to fall back to vmalloc when contiguous memory is unavailable
(instead of failing and logging page allocation warnings).

Signed-off-by: Carlos Bilbao (Lambda) <carlos.bilbao@kernel.org>
---
 drivers/infiniband/hw/irdma/utils.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/infiniband/hw/irdma/utils.c
b/drivers/infiniband/hw/irdma/utils.c
index 0422787592d8..59ef9856fd25 100644
--- a/drivers/infiniband/hw/irdma/utils.c
+++ b/drivers/infiniband/hw/irdma/utils.c
@@ -2257,7 +2257,7 @@ void irdma_pble_free_paged_mem(struct irdma_chunk
*chunk)
                                 chunk->pg_cnt);

 done:
-       kfree(chunk->dmainfo.dmaaddrs);
+       kvfree(chunk->dmainfo.dmaaddrs);
        chunk->dmainfo.dmaaddrs = NULL;
        vfree(chunk->vaddr);
        chunk->vaddr = NULL;
@@ -2274,7 +2274,7 @@ int irdma_pble_get_paged_mem(struct irdma_chunk
*chunk, u32 pg_cnt)
        u32 size;
        void *va;

-       chunk->dmainfo.dmaaddrs = kzalloc(pg_cnt << 3, GFP_KERNEL);
+       chunk->dmainfo.dmaaddrs = kvzalloc(pg_cnt << 3, GFP_KERNEL);
        if (!chunk->dmainfo.dmaaddrs)
                return -ENOMEM;

@@ -2295,7 +2295,7 @@ int irdma_pble_get_paged_mem(struct irdma_chunk
*chunk, u32 pg_cnt)

        return 0;
 err:
-       kfree(chunk->dmainfo.dmaaddrs);
+       kvfree(chunk->dmainfo.dmaaddrs);
        chunk->dmainfo.dmaaddrs = NULL;

        return -ENOMEM;
-- 2.50.1 (Apple Git-155)
Re: [PATCH] RDMA/irdma: Use kvzalloc for paged memory DMA address array
Posted by Leon Romanovsky 1 week, 5 days ago 
On Wed, Jan 21, 2026 at 10:54:36AM -0800, Carlos Bilbao wrote:
> Allocate array chunk->dmainfo.dmaaddrs using kvzalloc() to allow the
> allocation to fall back to vmalloc when contiguous memory is unavailable
> (instead of failing and logging page allocation warnings).
> 
> Signed-off-by: Carlos Bilbao (Lambda) <carlos.bilbao@kernel.org>
> ---
>  drivers/infiniband/hw/irdma/utils.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)

This patch doesn't apply to rdma/for-next.

Thanks
[PATCH v2] RDMA/irdma: Use kvzalloc for paged memory DMA address array
Posted by carlos.bilbao@kernel.org 1 week, 4 days ago 
From: Carlos Bilbao <carlos.bilbao@kernel.org>

Allocate array chunk->dmainfo.dmaaddrs using kvzalloc() to allow the
allocation to fall back to vmalloc when contiguous memory is unavailable
(instead of failing and logging page allocation warnings).

Acked-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com>
Signed-off-by: Carlos Bilbao (Lambda) <carlos.bilbao@kernel.org>
---
 drivers/infiniband/hw/irdma/utils.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/infiniband/hw/irdma/utils.c b/drivers/infiniband/hw/irdma/utils.c
index 13d7499131d4..6aac6d314564 100644
--- a/drivers/infiniband/hw/irdma/utils.c
+++ b/drivers/infiniband/hw/irdma/utils.c
@@ -2239,7 +2239,7 @@ void irdma_pble_free_paged_mem(struct irdma_chunk *chunk)
 				 chunk->pg_cnt);
 
 done:
-	kfree(chunk->dmainfo.dmaaddrs);
+	kvfree(chunk->dmainfo.dmaaddrs);
 	chunk->dmainfo.dmaaddrs = NULL;
 	vfree(chunk->vaddr);
 	chunk->vaddr = NULL;
@@ -2256,7 +2256,7 @@ int irdma_pble_get_paged_mem(struct irdma_chunk *chunk, u32 pg_cnt)
 	u32 size;
 	void *va;
 
-	chunk->dmainfo.dmaaddrs = kzalloc(pg_cnt << 3, GFP_KERNEL);
+	chunk->dmainfo.dmaaddrs = kvzalloc(pg_cnt << 3, GFP_KERNEL);
 	if (!chunk->dmainfo.dmaaddrs)
 		return -ENOMEM;
 
@@ -2277,7 +2277,7 @@ int irdma_pble_get_paged_mem(struct irdma_chunk *chunk, u32 pg_cnt)
 
 	return 0;
 err:
-	kfree(chunk->dmainfo.dmaaddrs);
+	kvfree(chunk->dmainfo.dmaaddrs);
 	chunk->dmainfo.dmaaddrs = NULL;
 
 	return -ENOMEM;
-- 
2.43.0
Re: [PATCH v2] RDMA/irdma: Use kvzalloc for paged memory DMA address array
Posted by Leon Romanovsky 1 week, 3 days ago 
On Tue, 27 Jan 2026 17:44:46 -0800, carlos.bilbao@kernel.org wrote:
> Allocate array chunk->dmainfo.dmaaddrs using kvzalloc() to allow the
> allocation to fall back to vmalloc when contiguous memory is unavailable
> (instead of failing and logging page allocation warnings).
> 
> 

Applied, thanks!

[1/1] RDMA/irdma: Use kvzalloc for paged memory DMA address array
      https://git.kernel.org/rdma/rdma/c/959d2c356e32ab

Best regards,
-- 
Leon Romanovsky <leon@kernel.org>
RE: [PATCH] RDMA/irdma: Use kvzalloc for paged memory DMA address array
Posted by Nikolova, Tatyana E 2 weeks, 2 days ago 

> -----Original Message-----
> From: Carlos Bilbao <carlos.bilbao@lambdal.com>
> Sent: Wednesday, January 21, 2026 12:55 PM
> To: mustafa.ismail@intel.com; shiraz.saleem@intel.com; jgg@ziepe.ca;
> leon@kernel.org
> Cc: bilbao@vt.edu; linux-rdma@vger.kernel.org; linux-kernel@vger.kernel.org;
> carlos.bilbao@kernel.org
> Subject: [PATCH] RDMA/irdma: Use kvzalloc for paged memory DMA address
> array
> 
> Allocate array chunk->dmainfo.dmaaddrs using kvzalloc() to allow the
> allocation to fall back to vmalloc when contiguous memory is unavailable
> (instead of failing and logging page allocation warnings).
> 
> Signed-off-by: Carlos Bilbao (Lambda) <carlos.bilbao@kernel.org>

Acked-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com>

> ---
>  drivers/infiniband/hw/irdma/utils.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/infiniband/hw/irdma/utils.c
> b/drivers/infiniband/hw/irdma/utils.c
> index 0422787592d8..59ef9856fd25 100644
> --- a/drivers/infiniband/hw/irdma/utils.c
> +++ b/drivers/infiniband/hw/irdma/utils.c
> @@ -2257,7 +2257,7 @@ void irdma_pble_free_paged_mem(struct
> irdma_chunk
> *chunk)
>                                  chunk->pg_cnt);
> 
>  done:
> -       kfree(chunk->dmainfo.dmaaddrs);
> +       kvfree(chunk->dmainfo.dmaaddrs);
>         chunk->dmainfo.dmaaddrs = NULL;
>         vfree(chunk->vaddr);
>         chunk->vaddr = NULL;
> @@ -2274,7 +2274,7 @@ int irdma_pble_get_paged_mem(struct
> irdma_chunk
> *chunk, u32 pg_cnt)
>         u32 size;
>         void *va;
> 
> -       chunk->dmainfo.dmaaddrs = kzalloc(pg_cnt << 3, GFP_KERNEL);
> +       chunk->dmainfo.dmaaddrs = kvzalloc(pg_cnt << 3, GFP_KERNEL);
>         if (!chunk->dmainfo.dmaaddrs)
>                 return -ENOMEM;
> 
> @@ -2295,7 +2295,7 @@ int irdma_pble_get_paged_mem(struct
> irdma_chunk
> *chunk, u32 pg_cnt)
> 
>         return 0;
>  err:
> -       kfree(chunk->dmainfo.dmaaddrs);
> +       kvfree(chunk->dmainfo.dmaaddrs);
>         chunk->dmainfo.dmaaddrs = NULL;
> 
>         return -ENOMEM;
> -- 2.50.1 (Apple Git-155)
> 
>
Re: [PATCH] RDMA/irdma: Use kvzalloc for paged memory DMA address array
Posted by Carlos Bilbao 2 weeks, 3 days ago 
Looks like Mustafa is no longer at Intel [1]. Does anyone know how
to reach them, or Shiraz? Their email addresses appear to be
bouncing as well.

Maybe you do, Andrew? 

[1] https://www.phoronix.com/news/Intel-More-Orphans-Maintainers

On 1/21/26 10:54 AM, Carlos Bilbao wrote:
> Allocate array chunk->dmainfo.dmaaddrs using kvzalloc() to allow the
> allocation to fall back to vmalloc when contiguous memory is unavailable
> (instead of failing and logging page allocation warnings).
>
> Signed-off-by: Carlos Bilbao (Lambda) <carlos.bilbao@kernel.org>
> ---
>  drivers/infiniband/hw/irdma/utils.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/infiniband/hw/irdma/utils.c
> b/drivers/infiniband/hw/irdma/utils.c
> index 0422787592d8..59ef9856fd25 100644
> --- a/drivers/infiniband/hw/irdma/utils.c
> +++ b/drivers/infiniband/hw/irdma/utils.c
> @@ -2257,7 +2257,7 @@ void irdma_pble_free_paged_mem(struct irdma_chunk
> *chunk)
>                                  chunk->pg_cnt);
>
>  done:
> -       kfree(chunk->dmainfo.dmaaddrs);
> +       kvfree(chunk->dmainfo.dmaaddrs);
>         chunk->dmainfo.dmaaddrs = NULL;
>         vfree(chunk->vaddr);
>         chunk->vaddr = NULL;
> @@ -2274,7 +2274,7 @@ int irdma_pble_get_paged_mem(struct irdma_chunk
> *chunk, u32 pg_cnt)
>         u32 size;
>         void *va;
>
> -       chunk->dmainfo.dmaaddrs = kzalloc(pg_cnt << 3, GFP_KERNEL);
> +       chunk->dmainfo.dmaaddrs = kvzalloc(pg_cnt << 3, GFP_KERNEL);
>         if (!chunk->dmainfo.dmaaddrs)
>                 return -ENOMEM;
>
> @@ -2295,7 +2295,7 @@ int irdma_pble_get_paged_mem(struct irdma_chunk
> *chunk, u32 pg_cnt)
>
>         return 0;
>  err:
> -       kfree(chunk->dmainfo.dmaaddrs);
> +       kvfree(chunk->dmainfo.dmaaddrs);
>         chunk->dmainfo.dmaaddrs = NULL;
>
>         return -ENOMEM;
> -- 2.50.1 (Apple Git-155)
>
>

Patchew 2.1-devel-b67e4c2

© 2016 - 2026 Red Hat, Inc.