This series changes the piix4_smbus driver's cd6h/cd7h port I/O accesses
to use MMIO instead. This is necessary because cd6h/cd7h port I/O may be
disabled on later AMD processors.
This series includes patches with MMIO accesses to register
FCH::PM::DECODEEN. The same register is also accessed by the sp5100_tco
driver.[1] Synchronization to the MMIO register is required in both
drivers.
The first patch creates a macro to request MMIO region using the 'muxed'
retry logic. This is used in patch 6 to synchronize accesses to EFCH MMIO.
The second patch replaces a hardcoded region size with a #define. This is
to improve maintainability and was requested from v2 review.
The third patch moves duplicated region request/release code into
functions. This locates related code into functions and reduces code line
count. This will also make adding MMIO support in patch 6 easier.
The fourth patch moves SMBus controller address detection into a function.
This is in preparation for adding MMIO region support.
The fifth patch moves EFCH port selection into a function. This is in
preparation for adding MMIO region support.
The sixth patch adds MMIO support for region requesting/releasing and
mapping. This is necessary for using MMIO to detect SMBus controller
address, enable SMBbus controller region, and control the port select.
The seventh patch updates the SMBus controller address detection to support
using MMIO. This is necessary because the driver accesses register
FCH::PM::DECODEEN during initialization and only available using MMIO on
later AMD processors.
The eighth patch updates the SMBus port selection to support MMIO. This is
required because port selection control resides in the
FCH::PM::DECODEEN[smbus0sel] and is only accessible using MMIO on later AMD
processors.
The ninth patch enables the EFCH MMIO functionality added earlier in this
series. The SMBus controller's PCI revision ID is used to check if EFCH
MMIO is supported by HW and should be enabled in the driver.
Based on v5.17-rc2.
Testing:
Tested on family 19h using:
i2cdetect -y 0
i2cdetect -y 2
- Results using v5.16 and this pachset applied. Below
shows the devices detected on the busses:
# i2cdetect -y 0
0 1 2 3 4 5 6 7 8 9 a b c d e f
00: -- -- -- -- -- -- -- --
10: 10 11 -- -- -- -- -- -- 18 -- -- -- -- -- -- --
20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
30: 30 -- -- -- -- 35 36 -- -- -- -- -- -- -- -- --
40: -- -- -- -- -- -- -- -- -- -- 4a -- -- -- -- --
50: 50 -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
60: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
70: -- -- -- 73 -- -- -- --
# i2cdetect -y 2
0 1 2 3 4 5 6 7 8 9 a b c d e f
00: -- -- -- -- -- -- -- --
10: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
30: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
40: -- -- -- -- -- -- -- -- -- -- -- -- 4c -- -- --
50: -- 51 -- -- 54 -- -- -- -- -- -- -- -- -- -- --
60: 60 -- -- 63 -- -- 66 -- -- -- -- 6b -- -- 6e --
70: 70 71 72 73 74 75 -- 77
Also tested using sp5100_tco submitted series listed below.[1]
I applied the sp5100_tco v4 series and ran:
cat >> /dev/watchdog
[1] sp5100_tco v4 patchset can be found here:
Link: https://lore.kernel.org/linux-watchdog/[email protected]/
Changes in v5:
- Use request/release helper function for sb800 device in
piix4_setup_sb800(). Patch 3. (Jean Delvare)
- Revert 'piix4_smba' variable definition ordering back as it was in
piix4_setup_sb800(). Patch 4. (Jean Delvare)
- Add newline after piix4_sb800_port_sel(). Patch 5. (Jean Delvare)
- Remove unnecessary initialization in piix4_add_adapter(). Patch 6.
(Jean Delvare)
- Remove unnecessary #define AMD_PCI_SMBUS_REVISION_MMIO. Patch 9.
(Jean Delvare)
- Add description for 0x51 constant moved in the above item. This is
in piix4_sb800_use_mmio(). Patch 9. (Andy Shevchenko)
- Rebase to v5.17-rc2. (Andy Shevchenko)
- Update patch 9 description. (Terry Bowman)
Changes in v4:
- Changed request_muxed_mem_region() macro to request_mem_region_muxed()
in patch 1. (Andy Shevchenko)
- Removed unnecessary newline where possible in calls to
request_muxed_region() in patch 2. (Terry Bowman)
- Changed piix4_sb800_region_setup() to piix4_sb800_region_request().
Patch 3. (Jean Delvare)
- Reordered piix4_setup_sb800() local variables from longest name length
to shortest name length. Patch 4. (Terry Bowman)
- Changed piix4_sb800_region_request() and piix4_sb800_region_release() by
adding early return() to remove 'else' improving readability. Patch 6.
(Terry Bowman)
- Removed iowrite32(ioread32(...), ...). Unnecessary because MMIO is
already enabled. (Terry Bowman)
- Refactored piix4_sb800_port_sel() to simplify the 'if' statement using
temp variable. Patch 8. (Terry Bowman)
- Added mmio_cfg.use_mmio assignment in piix4_add_adapter(). This is
needed for calls to piix4_sb800_port_sel() after initialization during
normal operation. Patch 9. (Terry Bowman)
Changes in v3:
- Added request_muxed_mem_region() patch (Wolfram, Guenter)
- Reduced To/Cc list length. (Andy)
Changes in v2:
- Split single patch. (Jean Delvare)
- Replace constant 2 with SB800_PIIX4_SMB_MAP_SIZE where appropriate.
(Jean Delvare)
- Shorten SB800_PIIX4_FCH_PM_DECODEEN_MMIO_EN name length to
SB800_PIIX4_FCH_PM_DECODEEN_MMIO. (Jean Delvare)
- Change AMD_PCI_SMBUS_REVISION_MMIO from 0x59 to 0x51. (Terry Bowman)
- Change piix4_sb800_region_setup() to piix4_sb800_region_request().
(Jean Delvare)
- Change 'SMB' text in logging to 'SMBus' (Jean Delvare)
- Remove unnecessary NULL assignment in piix4_sb800_region_release().
(Jean Delvare)
- Move 'u8' variable definitions to single line. (Jean Delvare)
- Hardcode piix4_setup_sb800_smba() return value to 0 since it is always
0. (Jean Delvare)
Terry Bowman (9):
kernel/resource: Introduce request_mem_region_muxed()
i2c: piix4: Replace hardcoded memory map size with a #define
i2c: piix4: Move port I/O region request/release code into functions
i2c: piix4: Move SMBus controller base address detect into function
i2c: piix4: Move SMBus port selection into function
i2c: piix4: Add EFCH MMIO support to region request and release
i2c: piix4: Add EFCH MMIO support to SMBus base address detect
i2c: piix4: Add EFCH MMIO support for SMBus port select
i2c: piix4: Enable EFCH MMIO for Family 17h+
drivers/i2c/busses/i2c-piix4.c | 213 ++++++++++++++++++++++++++-------
include/linux/ioport.h | 2 +
2 files changed, 169 insertions(+), 46 deletions(-)
--
2.30.2
Move duplicated region request and release code into a function. Move is
in preparation for following MMIO changes.
Signed-off-by: Terry Bowman <[email protected]>
Reviewed-by: Andy Shevchenko <[email protected]>
Reviewed-by: Jean Delvare <[email protected]>
---
drivers/i2c/busses/i2c-piix4.c | 48 ++++++++++++++++++++++------------
1 file changed, 31 insertions(+), 17 deletions(-)
diff --git a/drivers/i2c/busses/i2c-piix4.c b/drivers/i2c/busses/i2c-piix4.c
index 3ff68967034e..cc488b1e92c3 100644
--- a/drivers/i2c/busses/i2c-piix4.c
+++ b/drivers/i2c/busses/i2c-piix4.c
@@ -165,6 +165,24 @@ struct i2c_piix4_adapdata {
u8 port; /* Port number, shifted */
};
+static int piix4_sb800_region_request(struct device *dev)
+{
+ if (!request_muxed_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE,
+ "sb800_piix4_smb")) {
+ dev_err(dev,
+ "SMBus base address index region 0x%x already in use.\n",
+ SB800_PIIX4_SMB_IDX);
+ return -EBUSY;
+ }
+
+ return 0;
+}
+
+static void piix4_sb800_region_release(struct device *dev)
+{
+ release_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE);
+}
+
static int piix4_setup(struct pci_dev *PIIX4_dev,
const struct pci_device_id *id)
{
@@ -270,6 +288,7 @@ static int piix4_setup_sb800(struct pci_dev *PIIX4_dev,
unsigned short piix4_smba;
u8 smba_en_lo, smba_en_hi, smb_en, smb_en_status, port_sel;
u8 i2ccfg, i2ccfg_offset = 0x10;
+ int retval;
/* SB800 and later SMBus does not support forcing address */
if (force || force_addr) {
@@ -291,20 +310,16 @@ static int piix4_setup_sb800(struct pci_dev *PIIX4_dev,
else
smb_en = (aux) ? 0x28 : 0x2c;
- if (!request_muxed_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE,
- "sb800_piix4_smb")) {
- dev_err(&PIIX4_dev->dev,
- "SMB base address index region 0x%x already in use.\n",
- SB800_PIIX4_SMB_IDX);
- return -EBUSY;
- }
+ retval = piix4_sb800_region_request(&PIIX4_dev->dev);
+ if (retval)
+ return retval;
outb_p(smb_en, SB800_PIIX4_SMB_IDX);
smba_en_lo = inb_p(SB800_PIIX4_SMB_IDX + 1);
outb_p(smb_en + 1, SB800_PIIX4_SMB_IDX);
smba_en_hi = inb_p(SB800_PIIX4_SMB_IDX + 1);
- release_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE);
+ piix4_sb800_region_release(&PIIX4_dev->dev);
if (!smb_en) {
smb_en_status = smba_en_lo & 0x10;
@@ -373,11 +388,10 @@ static int piix4_setup_sb800(struct pci_dev *PIIX4_dev,
piix4_port_shift_sb800 = SB800_PIIX4_PORT_IDX_SHIFT;
}
} else {
- if (!request_muxed_region(SB800_PIIX4_SMB_IDX,
- SB800_PIIX4_SMB_MAP_SIZE,
- "sb800_piix4_smb")) {
+ retval = piix4_sb800_region_request(&PIIX4_dev->dev);
+ if (retval)
release_region(piix4_smba, SMBIOSIZE);
- return -EBUSY;
+ return retval;
}
outb_p(SB800_PIIX4_PORT_IDX_SEL, SB800_PIIX4_SMB_IDX);
@@ -387,7 +401,7 @@ static int piix4_setup_sb800(struct pci_dev *PIIX4_dev,
SB800_PIIX4_PORT_IDX;
piix4_port_mask_sb800 = SB800_PIIX4_PORT_IDX_MASK;
piix4_port_shift_sb800 = SB800_PIIX4_PORT_IDX_SHIFT;
- release_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE);
+ piix4_sb800_region_release(&PIIX4_dev->dev);
}
dev_info(&PIIX4_dev->dev,
@@ -685,9 +699,9 @@ static s32 piix4_access_sb800(struct i2c_adapter *adap, u16 addr,
u8 port;
int retval;
- if (!request_muxed_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE,
- "sb800_piix4_smb"))
- return -EBUSY;
+ retval = piix4_sb800_region_request(&adap->dev);
+ if (retval)
+ return retval;
/* Request the SMBUS semaphore, avoid conflicts with the IMC */
smbslvcnt = inb_p(SMBSLVCNT);
@@ -762,7 +776,7 @@ static s32 piix4_access_sb800(struct i2c_adapter *adap, u16 addr,
piix4_imc_wakeup();
release:
- release_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE);
+ piix4_sb800_region_release(&adap->dev);
return retval;
}
--
2.30.2
On Wed, Feb 09, 2022 at 11:27:08AM -0600, Terry Bowman wrote:
> This series changes the piix4_smbus driver's cd6h/cd7h port I/O accesses
> to use MMIO instead. This is necessary because cd6h/cd7h port I/O may be
> disabled on later AMD processors.
>
> This series includes patches with MMIO accesses to register
> FCH::PM::DECODEEN. The same register is also accessed by the sp5100_tco
> driver.[1] Synchronization to the MMIO register is required in both
> drivers.
>
> The first patch creates a macro to request MMIO region using the 'muxed'
> retry logic. This is used in patch 6 to synchronize accesses to EFCH MMIO.
>
> The second patch replaces a hardcoded region size with a #define. This is
> to improve maintainability and was requested from v2 review.
>
> The third patch moves duplicated region request/release code into
> functions. This locates related code into functions and reduces code line
> count. This will also make adding MMIO support in patch 6 easier.
>
> The fourth patch moves SMBus controller address detection into a function.
> This is in preparation for adding MMIO region support.
>
> The fifth patch moves EFCH port selection into a function. This is in
> preparation for adding MMIO region support.
>
> The sixth patch adds MMIO support for region requesting/releasing and
> mapping. This is necessary for using MMIO to detect SMBus controller
> address, enable SMBbus controller region, and control the port select.
>
> The seventh patch updates the SMBus controller address detection to support
> using MMIO. This is necessary because the driver accesses register
> FCH::PM::DECODEEN during initialization and only available using MMIO on
> later AMD processors.
>
> The eighth patch updates the SMBus port selection to support MMIO. This is
> required because port selection control resides in the
> FCH::PM::DECODEEN[smbus0sel] and is only accessible using MMIO on later AMD
> processors.
>
> The ninth patch enables the EFCH MMIO functionality added earlier in this
> series. The SMBus controller's PCI revision ID is used to check if EFCH
> MMIO is supported by HW and should be enabled in the driver.
>
> Based on v5.17-rc2.
>
> Testing:
> Tested on family 19h using:
> i2cdetect -y 0
> i2cdetect -y 2
>
> - Results using v5.16 and this pachset applied. Below
> shows the devices detected on the busses:
>
> # i2cdetect -y 0
> 0 1 2 3 4 5 6 7 8 9 a b c d e f
> 00: -- -- -- -- -- -- -- --
> 10: 10 11 -- -- -- -- -- -- 18 -- -- -- -- -- -- --
> 20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
> 30: 30 -- -- -- -- 35 36 -- -- -- -- -- -- -- -- --
> 40: -- -- -- -- -- -- -- -- -- -- 4a -- -- -- -- --
> 50: 50 -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
> 60: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
> 70: -- -- -- 73 -- -- -- --
> # i2cdetect -y 2
> 0 1 2 3 4 5 6 7 8 9 a b c d e f
> 00: -- -- -- -- -- -- -- --
> 10: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
> 20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
> 30: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
> 40: -- -- -- -- -- -- -- -- -- -- -- -- 4c -- -- --
> 50: -- 51 -- -- 54 -- -- -- -- -- -- -- -- -- -- --
> 60: 60 -- -- 63 -- -- 66 -- -- -- -- 6b -- -- 6e --
> 70: 70 71 72 73 74 75 -- 77
>
> Also tested using sp5100_tco submitted series listed below.[1]
> I applied the sp5100_tco v4 series and ran:
> cat >> /dev/watchdog
>
> [1] sp5100_tco v4 patchset can be found here:
> Link: https://lore.kernel.org/linux-watchdog/[email protected]/
>
> Changes in v5:
> - Use request/release helper function for sb800 device in
> piix4_setup_sb800(). Patch 3. (Jean Delvare)
> - Revert 'piix4_smba' variable definition ordering back as it was in
> piix4_setup_sb800(). Patch 4. (Jean Delvare)
> - Add newline after piix4_sb800_port_sel(). Patch 5. (Jean Delvare)
> - Remove unnecessary initialization in piix4_add_adapter(). Patch 6.
> (Jean Delvare)
> - Remove unnecessary #define AMD_PCI_SMBUS_REVISION_MMIO. Patch 9.
> (Jean Delvare)
> - Add description for 0x51 constant moved in the above item. This is
> in piix4_sb800_use_mmio(). Patch 9. (Andy Shevchenko)
> - Rebase to v5.17-rc2. (Andy Shevchenko)
> - Update patch 9 description. (Terry Bowman)
>
> Changes in v4:
> - Changed request_muxed_mem_region() macro to request_mem_region_muxed()
> in patch 1. (Andy Shevchenko)
> - Removed unnecessary newline where possible in calls to
> request_muxed_region() in patch 2. (Terry Bowman)
> - Changed piix4_sb800_region_setup() to piix4_sb800_region_request().
> Patch 3. (Jean Delvare)
> - Reordered piix4_setup_sb800() local variables from longest name length
> to shortest name length. Patch 4. (Terry Bowman)
> - Changed piix4_sb800_region_request() and piix4_sb800_region_release() by
> adding early return() to remove 'else' improving readability. Patch 6.
> (Terry Bowman)
> - Removed iowrite32(ioread32(...), ...). Unnecessary because MMIO is
> already enabled. (Terry Bowman)
> - Refactored piix4_sb800_port_sel() to simplify the 'if' statement using
> temp variable. Patch 8. (Terry Bowman)
> - Added mmio_cfg.use_mmio assignment in piix4_add_adapter(). This is
> needed for calls to piix4_sb800_port_sel() after initialization during
> normal operation. Patch 9. (Terry Bowman)
>
> Changes in v3:
> - Added request_muxed_mem_region() patch (Wolfram, Guenter)
> - Reduced To/Cc list length. (Andy)
>
> Changes in v2:
> - Split single patch. (Jean Delvare)
> - Replace constant 2 with SB800_PIIX4_SMB_MAP_SIZE where appropriate.
> (Jean Delvare)
> - Shorten SB800_PIIX4_FCH_PM_DECODEEN_MMIO_EN name length to
> SB800_PIIX4_FCH_PM_DECODEEN_MMIO. (Jean Delvare)
> - Change AMD_PCI_SMBUS_REVISION_MMIO from 0x59 to 0x51. (Terry Bowman)
> - Change piix4_sb800_region_setup() to piix4_sb800_region_request().
> (Jean Delvare)
> - Change 'SMB' text in logging to 'SMBus' (Jean Delvare)
> - Remove unnecessary NULL assignment in piix4_sb800_region_release().
> (Jean Delvare)
> - Move 'u8' variable definitions to single line. (Jean Delvare)
> - Hardcode piix4_setup_sb800_smba() return value to 0 since it is always
> 0. (Jean Delvare)
>
> Terry Bowman (9):
> kernel/resource: Introduce request_mem_region_muxed()
> i2c: piix4: Replace hardcoded memory map size with a #define
> i2c: piix4: Move port I/O region request/release code into functions
> i2c: piix4: Move SMBus controller base address detect into function
> i2c: piix4: Move SMBus port selection into function
> i2c: piix4: Add EFCH MMIO support to region request and release
> i2c: piix4: Add EFCH MMIO support to SMBus base address detect
> i2c: piix4: Add EFCH MMIO support for SMBus port select
> i2c: piix4: Enable EFCH MMIO for Family 17h+
>
Applied the series to for-next, thank you Terry for keeping at it and
Jean and Andy for the review. I'll send the pull request containing the
ioport update to the WDT maintainers now. All further changes should be
based on top of this now.
On Wed, 09 Feb 2022 11:27:11 -0600, Terry Bowman wrote:
> Move duplicated region request and release code into a function. Move is
> in preparation for following MMIO changes.
>
> Signed-off-by: Terry Bowman <[email protected]>
> Reviewed-by: Andy Shevchenko <[email protected]>
> Reviewed-by: Jean Delvare <[email protected]>
> ---
> drivers/i2c/busses/i2c-piix4.c | 48 ++++++++++++++++++++++------------
> 1 file changed, 31 insertions(+), 17 deletions(-)
>
> diff --git a/drivers/i2c/busses/i2c-piix4.c b/drivers/i2c/busses/i2c-piix4.c
> index 3ff68967034e..cc488b1e92c3 100644
> --- a/drivers/i2c/busses/i2c-piix4.c
> +++ b/drivers/i2c/busses/i2c-piix4.c
> @@ -165,6 +165,24 @@ struct i2c_piix4_adapdata {
> u8 port; /* Port number, shifted */
> };
>
> +static int piix4_sb800_region_request(struct device *dev)
> +{
> + if (!request_muxed_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE,
> + "sb800_piix4_smb")) {
> + dev_err(dev,
> + "SMBus base address index region 0x%x already in use.\n",
> + SB800_PIIX4_SMB_IDX);
> + return -EBUSY;
> + }
> +
> + return 0;
> +}
> +
> +static void piix4_sb800_region_release(struct device *dev)
> +{
> + release_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE);
> +}
> +
> static int piix4_setup(struct pci_dev *PIIX4_dev,
> const struct pci_device_id *id)
> {
> @@ -270,6 +288,7 @@ static int piix4_setup_sb800(struct pci_dev *PIIX4_dev,
> unsigned short piix4_smba;
> u8 smba_en_lo, smba_en_hi, smb_en, smb_en_status, port_sel;
> u8 i2ccfg, i2ccfg_offset = 0x10;
> + int retval;
>
> /* SB800 and later SMBus does not support forcing address */
> if (force || force_addr) {
> @@ -291,20 +310,16 @@ static int piix4_setup_sb800(struct pci_dev *PIIX4_dev,
> else
> smb_en = (aux) ? 0x28 : 0x2c;
>
> - if (!request_muxed_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE,
> - "sb800_piix4_smb")) {
> - dev_err(&PIIX4_dev->dev,
> - "SMB base address index region 0x%x already in use.\n",
> - SB800_PIIX4_SMB_IDX);
> - return -EBUSY;
> - }
> + retval = piix4_sb800_region_request(&PIIX4_dev->dev);
> + if (retval)
> + return retval;
>
> outb_p(smb_en, SB800_PIIX4_SMB_IDX);
> smba_en_lo = inb_p(SB800_PIIX4_SMB_IDX + 1);
> outb_p(smb_en + 1, SB800_PIIX4_SMB_IDX);
> smba_en_hi = inb_p(SB800_PIIX4_SMB_IDX + 1);
>
> - release_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE);
> + piix4_sb800_region_release(&PIIX4_dev->dev);
>
> if (!smb_en) {
> smb_en_status = smba_en_lo & 0x10;
> @@ -373,11 +388,10 @@ static int piix4_setup_sb800(struct pci_dev *PIIX4_dev,
> piix4_port_shift_sb800 = SB800_PIIX4_PORT_IDX_SHIFT;
> }
> } else {
> - if (!request_muxed_region(SB800_PIIX4_SMB_IDX,
> - SB800_PIIX4_SMB_MAP_SIZE,
> - "sb800_piix4_smb")) {
> + retval = piix4_sb800_region_request(&PIIX4_dev->dev);
> + if (retval)
Missing curly brace here, breaks the build.
> release_region(piix4_smba, SMBIOSIZE);
> - return -EBUSY;
> + return retval;
> }
>
> outb_p(SB800_PIIX4_PORT_IDX_SEL, SB800_PIIX4_SMB_IDX);
> @@ -387,7 +401,7 @@ static int piix4_setup_sb800(struct pci_dev *PIIX4_dev,
> SB800_PIIX4_PORT_IDX;
> piix4_port_mask_sb800 = SB800_PIIX4_PORT_IDX_MASK;
> piix4_port_shift_sb800 = SB800_PIIX4_PORT_IDX_SHIFT;
> - release_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE);
> + piix4_sb800_region_release(&PIIX4_dev->dev);
> }
>
> dev_info(&PIIX4_dev->dev,
> @@ -685,9 +699,9 @@ static s32 piix4_access_sb800(struct i2c_adapter *adap, u16 addr,
> u8 port;
> int retval;
>
> - if (!request_muxed_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE,
> - "sb800_piix4_smb"))
> - return -EBUSY;
> + retval = piix4_sb800_region_request(&adap->dev);
> + if (retval)
> + return retval;
>
> /* Request the SMBUS semaphore, avoid conflicts with the IMC */
> smbslvcnt = inb_p(SMBSLVCNT);
> @@ -762,7 +776,7 @@ static s32 piix4_access_sb800(struct i2c_adapter *adap, u16 addr,
> piix4_imc_wakeup();
>
> release:
> - release_region(SB800_PIIX4_SMB_IDX, SB800_PIIX4_SMB_MAP_SIZE);
> + piix4_sb800_region_release(&adap->dev);
> return retval;
> }
>
--
Jean Delvare
SUSE L3 Support
> > + if (retval)
>
> Missing curly brace here, breaks the build.
Bummer, need to check why this wasn't found by my build-testing.
I fixed it up and also rebased patch 6 to my change. Terry, please have
a look that I did everything correctly once I push out later today.
Thanks everyone!
On Wed, 09 Feb 2022 11:27:08 -0600, Terry Bowman wrote:
> This series changes the piix4_smbus driver's cd6h/cd7h port I/O accesses
> to use MMIO instead. This is necessary because cd6h/cd7h port I/O may be
> disabled on later AMD processors.
>
> This series includes patches with MMIO accesses to register
> FCH::PM::DECODEEN. The same register is also accessed by the sp5100_tco
> driver.[1] Synchronization to the MMIO register is required in both
> drivers.
> (...)
Except for the curly brace issue in patch 3, all looks good, so I
confirm my
Reviewed-by: Jean Delvare <[email protected]>
Tested-by: Jean Delvare <[email protected]>
on all i2c-piix4 patches.
Thanks,
--
Jean Delvare
SUSE L3 Support
Hi Wolfram and Jean,
On 2/11/22 08:42, Wolfram Sang wrote:
>>> + if (retval)
>>
>> Missing curly brace here, breaks the build.
>
> Bummer, need to check why this wasn't found by my build-testing.
>
> I fixed it up and also rebased patch 6 to my change. Terry, please have
> a look that I did everything correctly once I push out later today.
>
> Thanks everyone!
>
I need to look at how I sent that out. I will look for the fix in
i2c/for-next. Thanks for the help and review.
Hi Wolfram an Jean,
On 2/11/22 09:00, Terry Bowman wrote:
> Hi Wolfram and Jean,
>
> On 2/11/22 08:42, Wolfram Sang wrote:
>>>> + if (retval)
>>>
>>> Missing curly brace here, breaks the build.
>>
>> Bummer, need to check why this wasn't found by my build-testing.
>>
>> I fixed it up and also rebased patch 6 to my change. Terry, please have
>> a look that I did everything correctly once I push out later today.
>>
>> Thanks everyone!
>>
> I need to look at how I sent that out. I will look for the fix in
> i2c/for-next. Thanks for the help and review.
I verified the fix looks good. From commit a3325d225b00 on i2c/for-next:
@@ -373,11 +388,10 @@ static int piix4_setup_sb800(struct pci_dev *PIIX4_dev,
piix4_port_shift_sb800 = SB800_PIIX4_PORT_IDX_SHIFT;
}
} else {
- if (!request_muxed_region(SB800_PIIX4_SMB_IDX,
- SB800_PIIX4_SMB_MAP_SIZE,
- "sb800_piix4_smb")) {
+ retval = piix4_sb800_region_request(&PIIX4_dev->dev);
+ if (retval) {
release_region(piix4_smba, SMBIOSIZE);
- return -EBUSY;
+ return retval;
}
> Maybe you build-tested the series as a whole but not individual
> patches? The series did build fine, as the missing curly brace was
> added back in a later patch.
It wasn't that, but another gory detail. Updating my build tests is in
the works. But thanks for the help!
On Fri, 11 Feb 2022 15:42:34 +0100, Wolfram Sang wrote:
> > > + if (retval)
> >
> > Missing curly brace here, breaks the build.
>
> Bummer, need to check why this wasn't found by my build-testing.
Maybe you build-tested the series as a whole but not individual
patches? The series did build fine, as the missing curly brace was
added back in a later patch.
--
Jean Delvare
SUSE L3 Support