* [U-Boot] [PATCH v2] ata: ahci allow 64-bit DMA for SATA
@ 2019-08-28 9:24 Roman Kapl
2019-10-11 14:41 ` Tom Rini
0 siblings, 1 reply; 2+ messages in thread
From: Roman Kapl @ 2019-08-28 9:24 UTC (permalink / raw)
To: u-boot
Allow 64-bit DMA on AHCI. If not supported by the host controller, at
least print a message and fail.
Signed-off-by: Roman Kapl <rka@sysgo.com>
---
Please disregard the previous patch, I've send a wrong version that does not
even compile.
drivers/ata/ahci.c | 14 +++++++++++---
1 file changed, 11 insertions(+), 3 deletions(-)
diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
index 9a08575053..fd4df60a17 100644
--- a/drivers/ata/ahci.c
+++ b/drivers/ata/ahci.c
@@ -50,6 +50,8 @@ struct ahci_uc_priv *probe_ent = NULL;
#define WAIT_MS_FLUSH 5000
#define WAIT_MS_LINKUP 200
+#define AHCI_CAP_S64A (1u << 31)
+
__weak void __iomem *ahci_port_base(void __iomem *base, u32 port)
{
return base + 0x100 + (port * 0x80);
@@ -503,9 +505,15 @@ static int ahci_fill_sg(struct ahci_uc_priv *uc_priv, u8 port,
}
for (i = 0; i < sg_count; i++) {
- ahci_sg->addr =
- cpu_to_le32((unsigned long) buf + i * MAX_DATA_BYTE_COUNT);
- ahci_sg->addr_hi = 0;
+ /* We assume virt=phys */
+ phys_addr_t pa = (unsigned long)buf + i * MAX_DATA_BYTE_COUNT;
+
+ ahci_sg->addr = cpu_to_le32(pa & U32_MAX);
+ ahci_sg->addr_hi = cpu_to_le32((pa >> 32) & U32_MAX);
+ if (ahci_sg->addr_hi && !(uc_priv->cap & AHCI_CAP_S64A)) {
+ printf("Error: DMA address too high\n");
+ return -1;
+ }
ahci_sg->flags_size = cpu_to_le32(0x3fffff &
(buf_len < MAX_DATA_BYTE_COUNT
? (buf_len - 1)
--
2.22.0
^ permalink raw reply related [flat|nested] 2+ messages in thread
* [U-Boot] [PATCH v2] ata: ahci allow 64-bit DMA for SATA
2019-08-28 9:24 [U-Boot] [PATCH v2] ata: ahci allow 64-bit DMA for SATA Roman Kapl
@ 2019-10-11 14:41 ` Tom Rini
0 siblings, 0 replies; 2+ messages in thread
From: Tom Rini @ 2019-10-11 14:41 UTC (permalink / raw)
To: u-boot
On Wed, Aug 28, 2019 at 11:24:35AM +0200, Roman Kapl wrote:
> Allow 64-bit DMA on AHCI. If not supported by the host controller, at
> least print a message and fail.
>
> Signed-off-by: Roman Kapl <rka@sysgo.com>
> ---
>
> Please disregard the previous patch, I've send a wrong version that does not
> even compile.
>
> drivers/ata/ahci.c | 14 +++++++++++---
> 1 file changed, 11 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
> index 9a08575053..fd4df60a17 100644
> --- a/drivers/ata/ahci.c
> +++ b/drivers/ata/ahci.c
> @@ -50,6 +50,8 @@ struct ahci_uc_priv *probe_ent = NULL;
> #define WAIT_MS_FLUSH 5000
> #define WAIT_MS_LINKUP 200
>
> +#define AHCI_CAP_S64A (1u << 31)
> +
> __weak void __iomem *ahci_port_base(void __iomem *base, u32 port)
> {
> return base + 0x100 + (port * 0x80);
> @@ -503,9 +505,15 @@ static int ahci_fill_sg(struct ahci_uc_priv *uc_priv, u8 port,
> }
>
> for (i = 0; i < sg_count; i++) {
> - ahci_sg->addr =
> - cpu_to_le32((unsigned long) buf + i * MAX_DATA_BYTE_COUNT);
> - ahci_sg->addr_hi = 0;
> + /* We assume virt=phys */
> + phys_addr_t pa = (unsigned long)buf + i * MAX_DATA_BYTE_COUNT;
> +
> + ahci_sg->addr = cpu_to_le32(pa & U32_MAX);
> + ahci_sg->addr_hi = cpu_to_le32((pa >> 32) & U32_MAX);
> + if (ahci_sg->addr_hi && !(uc_priv->cap & AHCI_CAP_S64A)) {
> + printf("Error: DMA address too high\n");
> + return -1;
> + }
> ahci_sg->flags_size = cpu_to_le32(0x3fffff &
> (buf_len < MAX_DATA_BYTE_COUNT
> ? (buf_len - 1)
This version breaks building on qemu_arm:
arm: + qemu_arm
+(qemu_arm) In file included from arch/arm/include/asm/byteorder.h:29:0,
+(qemu_arm) from include/compiler.h:128,
+(qemu_arm) from include/image.h:18,
+(qemu_arm) from include/common.h:41,
+(qemu_arm) from drivers/ata/ahci.c:11:
+(qemu_arm) drivers/ata/ahci.c: In function 'ahci_fill_sg':
+(qemu_arm) drivers/ata/ahci.c:512:38: error: right shift count >= width of type [-Werror=shift-count-overflow]
+(qemu_arm) ahci_sg->addr_hi = cpu_to_le32((pa >> 32) & U32_MAX);
+(qemu_arm) ^
+(qemu_arm) include/linux/byteorder/little_endian.h:34:51: note: in definition of macro '__cpu_to_le32'
+(qemu_arm) #define __cpu_to_le32(x) ((__force __le32)(__u32)(x))
+(qemu_arm) ^
+(qemu_arm) drivers/ata/ahci.c:512:22: note: in expansion of macro 'cpu_to_le32'
+(qemu_arm) ^~~~~~~~~~~
+(qemu_arm) cc1: all warnings being treated as errors
+(qemu_arm) make[3]: *** [drivers/ata/ahci.o] Error 1
+(qemu_arm) make[2]: *** [drivers/ata] Error 2
+(qemu_arm) make[1]: *** [drivers] Error 2
+(qemu_arm) make: *** [sub-make] Error 2
--
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20191011/74fa2ef8/attachment.sig>
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2019-10-11 14:41 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-08-28 9:24 [U-Boot] [PATCH v2] ata: ahci allow 64-bit DMA for SATA Roman Kapl
2019-10-11 14:41 ` Tom Rini
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.