hw/sd/sdcard: Cleanups before adding eMMC support
v45 cover hw/sd/sdcard: Cleanups before adding eMMC support
(patches from v42 already reviewed not reposted)
- Addressed review comments from Daniel & Luc wrt migration
- Remove old comment
Philippe Mathieu-Daudé (3):
hw/sd/sdcard: Remove leftover comment about removed 'spi' Property hw/sd/sdcard: Use spec v3.01 by default
hw/sd/sdcard: Do not store vendor data on block drive (CMD56)
hw/core/machine.c | 1 + hw/sd/sd.c | 35 ++++++++++++++++++++--------------- 2 files changed, 21 insertions(+), 15 deletions(-)
Signature · 1 line
2.41.0
PATCH v45 1/3 hw/sd/sdcard: Remove leftover comment about removed 'spi' Property
Commit c3287c0f70 ("hw/sd: Introduce a "sd-card" SPI variant model") removed the 'spi' property. Remove the comment left over.
--- hw/sd/sd.c | 4 ---- 1 file changed, 4 deletions(-)
hw/sd/sd.c +0 -4
@@ -2473,10 +2473,6 @@ static Property sd_properties[] = {DEFINE_PROP_UINT8 ("spec_version" , SDState, spec_version, SD_PHY_SPECv2_00_VERS),DEFINE_PROP_DRIVE ("drive" , SDState, blk), -/* We do not model the chip select pin, so allow the board to select -* whether card should be in SSI or MMC/SD mode. It is also up to the -* board to ensure that ssi transfers only occur when the chip select -* is asserted. */ DEFINE_PROP_END_OF_LIST() };
Signature · 1 line
2.41.0
Re: [PATCH v45 1/3] hw/sd/sdcard: Remove leftover comment about removed 'spi' Property
Quoted 28 lines · Philippe Mathieu-Daudé
Commit
c3287c0f70("hw/sd: Introduce a "sd-card" SPI variant model") removed the 'spi' property. Remove the comment left over.Signed-off-by Philippe Mathieu-Daudé <[email protected]>---
hw/sd/sd.c| 4 ----1 file changed, 4 deletions(-)hw/sd/sd.c +0
-4@@ -2473,10 +2473,6 @@ static Property sd_properties[] = {DEFINE_PROP_UINT8 ("spec_version" , SDState, spec_version, SD_PHY_SPECv2_00_VERS),DEFINE_PROP_DRIVE ("drive" , SDState, blk), -/* We do not model the chip select pin, so allow the board to select -* whether card should be in SSI or MMC/SD mode. It is also up to the -* board to ensure that ssi transfers only occur when the chip select -* is asserted. */ DEFINE_PROP_END_OF_LIST() };Signature · 1 line
2.41.0
PATCH v45 2/3 hw/sd/sdcard: Use spec v3.01 by default
In reply to Philippe Mathieu-DaudéRecent SDHCI expect cards to support the v3.01 spec to negociate lower I/O voltage. Select it by default.
Versioned machine types with a version of 9.0 or earlier retain the old default (spec v2.00).
---
v43: update versioned machines (danpb)
--- hw/core/machine.c | 1 + hw/sd/sd.c | 2 +- 2 files changed, 2 insertions(+), 1 deletion(-)
hw/core/machine.c +1 -0
@@ -38,6 +38,7 @@ GlobalProperty hw_compat_9_0[] = { {"arm-cpu", "backcompat-cntfrq", "true" }, {"scsi-disk-base", "migrate-emulated-scsi-request", "false" }, {"vfio-pci", "skip-vsc-check", "false" }, + {"sd-card", "spec_version", "2" }, };const size_t hw_compat_9_0_len =G_N_ELEMENTS (hw_compat_9_0);
hw/sd/sd.c +1 -1
@@ -2471,7 +2471,7 @@ static void sd_realize(DeviceState *dev, Error **errp)static Property sd_properties[] = {DEFINE_PROP_UINT8 ("spec_version" , SDState, - spec_version, SD_PHY_SPECv2_00_VERS), + spec_version, SD_PHY_SPECv3_01_VERS),DEFINE_PROP_DRIVE ("drive" , SDState, blk),DEFINE_PROP_END_OF_LIST () };
Signature · 1 line
2.41.0
PATCH v45 3/3 hw/sd/sdcard: Do not store vendor data on block drive (CMD56)
In reply to Philippe Mathieu-Daudé"General command" (GEN_CMD, CMD56) is described as:
GEN_CMD is the same as the single block read or write commands (CMD24 or CMD17). The difference is that [...] the data block is not a memory payload data but has a vendor specific format and meaning.
Thus this block must not be stored overwriting data block on underlying storage drive. Keep it in a dedicated 'vendor_data[]' array.
---
v43: Do not re-use VMSTATE_UNUSED_V (danpb)
v44: Use subsection (Luc)
v45: Remove APP_READ_BLOCK/APP_WRITE_BLOCK macros
--- hw/sd/sd.c | 29 +++++++++++++++++++---------- 1 file changed, 19 insertions(+), 10 deletions(-)
hw/sd/sd.c +19 -10
@@ -153,6 +153,8 @@ struct SDState {uint32_t data_offset;size_t data_size;uint8_t data[512 ]; +uint8_t vendor_data[512 ]; +qemu_irq readonly_cb;qemu_irq inserted_cb;QEMUTimer *ocr_power_timer;
@@ -719,6 +721,7 @@ static void sd_reset(DeviceState *dev) sd->wp_switch = sd->blk ? !blk_is_writable (sd->blk) : false; sd->wp_group_bits = sect; sd->wp_group_bmap =bitmap_new (sd->wp_group_bits); +memset (sd->vendor_data,0xec ,sizeof (sd->vendor_data));memset (sd->function_group,0 ,sizeof (sd->function_group)); sd->erase_start =INVALID_ADDRESS ; sd->erase_end =INVALID_ADDRESS ;
@@ -793,6 +796,16 @@ static const VMStateDescription sd_ocr_vmstate = { }, }; +static const VMStateDescription sd_vendordata_vmstate = { + .name ="sd-card/vendor_data-state" , + .version_id =1 , + .minimum_version_id =1 , + .fields = (const VMStateField []) { +VMSTATE_UINT8_ARRAY (vendor_data, SDState,512 ), +VMSTATE_END_OF_LIST () + }, +}; +static int sd_vmstate_pre_load (void *opaque) {SDState *sd = opaque;
@@ -840,6 +853,7 @@ static const VMStateDescription sd_vmstate = { }, .subsections = (const VMStateDescription * const []) { &sd_ocr_vmstate, + &sd_vendordata_vmstate,NULL }, };
@@ -902,9 +916,6 @@ static void sd_blk_write(SDState *sd, uint64_t addr, uint32_t len) } } -#define APP_READ_BLOCK (a, len) memset(sd->data, 0xec, len) -#define APP_WRITE_BLOCK (a, len) -static void sd_erase (SDState *sd) {uint64_t erase_start = sd->erase_start;
@@ -2187,9 +2198,8 @@ void sd_write_byte(SDState *sd, uint8_t value)break ;case 56 :/* CMD56: GEN_CMD */ - sd->data[sd->data_offset ++] = value; -if (sd->data_offset >= sd->blk_len) { -APP_WRITE_BLOCK (sd->data_start, sd->data_offset); + sd->vendor_data[sd->data_offset++] = value; +if (sd->data_offset >=sizeof (sd->vendor_data)) { sd->state = sd_transfer_state; }break ;
@@ -2261,12 +2271,11 @@ uint8_t sd_read_byte(SDState *sd)break ;case 56 :/* CMD56: GEN_CMD */ -if (sd->data_offset ==0 ) -APP_READ_BLOCK (sd->data_start, sd->blk_len); - ret = sd->data[sd->data_offset ++]; + ret = sd->vendor_data[sd->data_offset++]; -if (sd->data_offset >= sd->blk_len) +if (sd->data_offset >=sizeof (sd->vendor_data)) { sd->state = sd_transfer_state; + }break ;default :
Signature · 1 line
2.41.0
Re: [PATCH v45 3/3] hw/sd/sdcard: Do not store vendor data on block drive (CMD56)
Quoted 11 lines · Philippe Mathieu-Daudé
"General command" (
GEN_CMD, CMD56) is described as:
GEN_CMDis the same as the single block read or write commands (CMD24 or CMD17). The difference is that [...] the data block is not a memory payload data but has a vendor specific format and meaning.Thus this block must not be stored overwriting data block on underlying storage drive. Keep it in a dedicated 'vendor_data[]' array.
I am reading the 4.3 spec, and it says:
The bus transaction of the GEN_CMD is the same as the single block read or write commands (CMD24 or CMD17). The difference is that the argument denotes the direction of the data transfer (rather than the address) and the data block is not a memory payload data but has a vendor specific format and meaning.
The vendor here (qemu) does not support any functionality with CMD56. I think the correct approach would be to read zeros and discard writes, without storing anything and without changing data_offset (which is for data buffer)
What do you think?
Quoted 97 lines
Signed-off-by Philippe Mathieu-Daudé <[email protected]>---
v43: Do not re-useVMSTATE_UNUSED_V(danpb)
v44: Use subsection (Luc)
v45: Remove APP_READ_BLOCK/APP_WRITE_BLOCK macros
---hw/sd/sd.c| 29 +++++++++++++++++++----------1 file changed, 19 insertions(+), 10 deletions(-)hw/sd/sd.c +19
-10@@ -153,6 +153,8 @@ struct SDState {uint32_t data_offset;size_t data_size;uint8_t data[512 ]; +uint8_t vendor_data[512 ]; +qemu_irq readonly_cb;qemu_irq inserted_cb;QEMUTimer *ocr_power_timer;@@ -719,6 +721,7 @@ static void sd_reset(DeviceState *dev) sd->wp_switch = sd->blk ? !blk_is_writable (sd->blk) : false; sd->wp_group_bits = sect; sd->wp_group_bmap =bitmap_new (sd->wp_group_bits); +memset (sd->vendor_data,0xec ,sizeof (sd->vendor_data));memset (sd->function_group,0 ,sizeof (sd->function_group)); sd->erase_start =INVALID_ADDRESS ; sd->erase_end =INVALID_ADDRESS ;@@ -793,6 +796,16 @@ static const VMStateDescription sd_ocr_vmstate = { }, }; +static const VMStateDescription sd_vendordata_vmstate = { + .name ="sd-card/vendor_data-state" , + .version_id =1 , + .minimum_version_id =1 , + .fields = (const VMStateField []) { +VMSTATE_UINT8_ARRAY (vendor_data, SDState,512 ), +VMSTATE_END_OF_LIST () + }, +}; +static int sd_vmstate_pre_load (void *opaque) {SDState *sd = opaque;@@ -840,6 +853,7 @@ static const VMStateDescription sd_vmstate = { }, .subsections = (const VMStateDescription * const []) { &sd_ocr_vmstate, + &sd_vendordata_vmstate,NULL }, };@@ -902,9 +916,6 @@ static void sd_blk_write(SDState *sd, uint64_t addr, uint32_t len) } } -#define APP_READ_BLOCK (a, len) memset(sd->data, 0xec, len) -#define APP_WRITE_BLOCK (a, len) -static void sd_erase (SDState *sd) {uint64_t erase_start = sd->erase_start;@@ -2187,9 +2198,8 @@ void sd_write_byte(SDState *sd, uint8_t value)break ;case 56 :/* CMD56: GEN_CMD */ - sd->data[sd->data_offset ++] = value; -if (sd->data_offset >= sd->blk_len) { -APP_WRITE_BLOCK (sd->data_start, sd->data_offset); + sd->vendor_data[sd->data_offset++] = value; +if (sd->data_offset >=sizeof (sd->vendor_data)) { sd->state = sd_transfer_state; }break ;@@ -2261,12 +2271,11 @@ uint8_t sd_read_byte(SDState *sd)break ;case 56 :/* CMD56: GEN_CMD */ -if (sd->data_offset ==0 ) -APP_READ_BLOCK (sd->data_start, sd->blk_len); - ret = sd->data[sd->data_offset ++]; + ret = sd->vendor_data[sd->data_offset++]; -if (sd->data_offset >= sd->blk_len) +if (sd->data_offset >=sizeof (sd->vendor_data)) { sd->state = sd_transfer_state; + }break ;default :Signature · 1 line
2.41.0
Re: [PATCH v45 3/3] hw/sd/sdcard: Do not store vendor data on block drive (CMD56)
Quoted 22 lines · Manos Pitsidianakis
Quoted 11 lines · Philippe Mathieu-Daudé
"General command" (
GEN_CMD, CMD56) is described as:GEN_CMD is the same as the single block read or write commands (CMD24 or CMD17). The difference is that [...] the data block is not a memory payload data but has a vendor specific format and meaning.
Thus this block must not be stored overwriting data block on underlying storage drive. Keep it in a dedicated 'vendor_data[]' array.
I am reading the 4.3 spec, and it says:
The bus transaction of the
GEN_CMDis the same as the single block read or write commands (CMD24 or CMD17). The difference is that the argument denotes the direction of the data transfer (rather than the address) and the data block is not a memory payload data but has a vendor specific format and meaning.
Quoted 3 lines
The vendor here (qemu) does not support any functionality with CMD56. I think the correct approach would be to read zeros and discard writes, without storing anything and without changing
data_offset(which is fordatabuffer)What do you think?
As Luc suggested in v42. Indeed simpler thus clever.
Quoted 8 lines
Quoted 8 lines
Signed-off-by Philippe Mathieu-Daudé <[email protected]>---
v43: Do not re-useVMSTATE_UNUSED_V(danpb)
v44: Use subsection (Luc)
v45: Remove APP_READ_BLOCK/APP_WRITE_BLOCK macros
---hw/sd/sd.c| 29 +++++++++++++++++++----------1 file changed, 19 insertions(+), 10 deletions(-)