Re: [PATCH v2] firmware: gsmi: Drop the use of dma_pool_* API functions

From: Ard Biesheuvel
Date: Thu Oct 22 2020 - 03:02:42 EST


On Thu, 22 Oct 2020 at 06:38, Furquan Shaikh <furquan@xxxxxxxxxx> wrote:
>
> GSMI driver uses dma_pool_* API functions for buffer allocation
> because it requires that the SMI buffers are allocated within 32-bit
> physical address space. However, this does not work well with IOMMU
> since there is no real device and hence no domain associated with the
> device.
>
> Since this is not a real device, it does not require any device
> address(IOVA) for the buffer allocations. The only requirement is to
> ensure that the physical address allocated to the buffer is within
> 32-bit physical address space. This is because the buffers have
> nothing to do with DMA at all. It is required for communication with
> firmware executing in SMI mode which has access only to the bottom
> 4GiB of memory. Hence, this change switches to using a SLAB cache
> created with SLAB_CACHE_DMA32 that guarantees that the allocation
> happens from the DMA32 memory zone. All calls to dma_pool_* are
> replaced with kmem_cache_*.
>
> In addition to that, all the code for managing the dma_pool for GSMI
> platform device is dropped.
>
> Signed-off-by: Furquan Shaikh <furquan@xxxxxxxxxx>
> ---
> Changelog since v1:
> - Switched to using SLAB cache with SLAB_CACHE_DMA32.
> - Added comment to code and commit message explaining the reason for
> using DMA32 memory zone.
>
> drivers/firmware/google/gsmi.c | 38 +++++++++++++++++++++++-----------
> 1 file changed, 26 insertions(+), 12 deletions(-)
>
> diff --git a/drivers/firmware/google/gsmi.c b/drivers/firmware/google/gsmi.c
> index 7d9367b22010..092d8cb209a3 100644
> --- a/drivers/firmware/google/gsmi.c
> +++ b/drivers/firmware/google/gsmi.c
> @@ -17,7 +17,6 @@
> #include <linux/string.h>
> #include <linux/spinlock.h>
> #include <linux/dma-mapping.h>
> -#include <linux/dmapool.h>
> #include <linux/fs.h>
> #include <linux/slab.h>
> #include <linux/ioctl.h>
> @@ -85,7 +84,6 @@
> struct gsmi_buf {
> u8 *start; /* start of buffer */
> size_t length; /* length of buffer */
> - dma_addr_t handle; /* dma allocation handle */
> u32 address; /* physical address of buffer */
> };
>
> @@ -97,7 +95,7 @@ static struct gsmi_device {
> spinlock_t lock; /* serialize access to SMIs */
> u16 smi_cmd; /* SMI command port */
> int handshake_type; /* firmware handler interlock type */
> - struct dma_pool *dma_pool; /* DMA buffer pool */
> + struct kmem_cache *mem_pool; /* kmem cache for gsmi_buf allocations */
> } gsmi_dev;
>
> /* Packed structures for communicating with the firmware */
> @@ -157,14 +155,20 @@ static struct gsmi_buf *gsmi_buf_alloc(void)
> }
>
> /* allocate buffer in 32bit address space */
> - smibuf->start = dma_pool_alloc(gsmi_dev.dma_pool, GFP_KERNEL,
> - &smibuf->handle);
> + smibuf->start = kmem_cache_alloc(gsmi_dev.mem_pool, GFP_KERNEL);
> if (!smibuf->start) {
> printk(KERN_ERR "gsmi: failed to allocate name buffer\n");
> kfree(smibuf);
> return NULL;
> }
>
> + if ((u64)virt_to_phys(smibuf->start) >> 32) {
> + printk(KERN_ERR "gsmi: allocation not within 32-bit physical address space\n");
> + kfree(smibuf->start);
> + kfree(smibuf);
> + return NULL;
> + }
> +

Please drop this check - on x86, DMA32 guarantees that the buffer is
in 32-bit address space.

With the check dropped:

Reviewed-by: Ard Biesheuvel <ardb@xxxxxxxxxx>


> /* fill in the buffer handle */
> smibuf->length = GSMI_BUF_SIZE;
> smibuf->address = (u32)virt_to_phys(smibuf->start);
> @@ -176,8 +180,7 @@ static void gsmi_buf_free(struct gsmi_buf *smibuf)
> {
> if (smibuf) {
> if (smibuf->start)
> - dma_pool_free(gsmi_dev.dma_pool, smibuf->start,
> - smibuf->handle);
> + kmem_cache_free(gsmi_dev.mem_pool, smibuf->start);
> kfree(smibuf);
> }
> }
> @@ -914,9 +917,20 @@ static __init int gsmi_init(void)
> spin_lock_init(&gsmi_dev.lock);
>
> ret = -ENOMEM;
> - gsmi_dev.dma_pool = dma_pool_create("gsmi", &gsmi_dev.pdev->dev,
> - GSMI_BUF_SIZE, GSMI_BUF_ALIGN, 0);
> - if (!gsmi_dev.dma_pool)
> +
> + /*
> + * SLAB cache is created using SLAB_CACHE_DMA32 to ensure that the
> + * allocations for gsmi_buf come from the DMA32 memory zone. These
> + * buffers have nothing to do with DMA. They are required for
> + * communication with firmware executing in SMI mode which can only
> + * access the bottom 4GiB of physical memory. Since DMA32 memory zone
> + * guarantees allocation under the 4GiB boundary, this driver creates
> + * a SLAB cache with SLAB_CACHE_DMA32 flag.
> + */
> + gsmi_dev.mem_pool = kmem_cache_create("gsmi", GSMI_BUF_SIZE,
> + GSMI_BUF_ALIGN,
> + SLAB_CACHE_DMA32, NULL);
> + if (!gsmi_dev.mem_pool)
> goto out_err;
>
> /*
> @@ -1032,7 +1046,7 @@ static __init int gsmi_init(void)
> gsmi_buf_free(gsmi_dev.param_buf);
> gsmi_buf_free(gsmi_dev.data_buf);
> gsmi_buf_free(gsmi_dev.name_buf);
> - dma_pool_destroy(gsmi_dev.dma_pool);
> + kmem_cache_destroy(gsmi_dev.mem_pool);
> platform_device_unregister(gsmi_dev.pdev);
> pr_info("gsmi: failed to load: %d\n", ret);
> #ifdef CONFIG_PM
> @@ -1057,7 +1071,7 @@ static void __exit gsmi_exit(void)
> gsmi_buf_free(gsmi_dev.param_buf);
> gsmi_buf_free(gsmi_dev.data_buf);
> gsmi_buf_free(gsmi_dev.name_buf);
> - dma_pool_destroy(gsmi_dev.dma_pool);
> + kmem_cache_destroy(gsmi_dev.mem_pool);
> platform_device_unregister(gsmi_dev.pdev);
> #ifdef CONFIG_PM
> platform_driver_unregister(&gsmi_driver_info);
> --
> 2.29.0.rc1.297.gfa9743e501-goog
>