Skip to contentSkip to conversation

sdhci: fix SDMA boundary bug

PATCH Superseded Tao Ding 2 participants 7 messages Jun 17 – Jun 28

v1 of 7 Superseded Latest is v7 →

lore
Share & tools
lore

Tao Ding

author
#0 lore raw

cover sdhci: fix SDMA boundary bug

Hi all,

Booting Linux from SD card on Xilinx Zynq with SDMA (MMC_SDHCI_SDMA=y) currently hangs during image load.

This patch fixes the hang and enables successful image loading.

Thanks!

Tao Ding (1):
fixed sdma boundary bug

 hw/sd/sdhci.c | 2 ++
 1 file changed, 2 insertions(+)

base-commit: 608ad9307d5b4797d2254ee3a4bad2bc8c8bcbae

Signature · 1 line
2.43.0
author
#1 lore raw

PATCH 1/1 fixed sdma boundary bug

During SDMA transfers, the controller raises an interrupt at buffer boundaries to request a system address update. But the emulation fails to clear the SDHC_DOING_READ/WRITE flags after DMA completion, so writes to SDHC_SYSAD are ignored. Clear these flags when the transfer ends.
(according to PartA2_SD_Host_Controller_Simplified_Specification_Ver2.00.pdf section 2.2.1)

This is a QTest in npcm7xx_sdhci-test that exercises this boundary condition and validates the fix.

QTest steps:
1. Replace npcm7xx_sdhci-test.c with the updated version (full code available at: https://pastebin.com/AWVp2wFi).
2. Build with:

       ../configure --target-list="arm-softmmu,aarch64-softmmu" \
                    --disable-slirp --enable-debug --disable-werror
       ninja qemu-system-arm tests/qtest/npcm7xx_sdhci-test
  3. Run the specific test case:
       QTEST_QEMU_BINARY=./qemu-system-arm \
         ./tests/qtest/npcm7xx_sdhci-test \
         -p /arm/npcm7xx_sdhci/read_sd_sdma_boundary_continue

Before the fix, the test fails with a BLKCNT assertion:
ERROR:...wait_for_block_count: assertion failed (qtest_readw(...) == expected): (1 == 0)

After applying the fix, the test passes:
ok 1 /arm/npcm7xx_sdhci/read_sd_sdma_boundary_continue

The new test case ensures the SDMA boundary handling works correctly and prevents future regressions.

Uboot test:
1. Prepare zImage, uboot and rootfs.cpio.gz, make sure uboot config (MMC_SDHCI_SDMA=y).
2. ./qemu-system-aarch64 -M xilinx-zynq-a9 -machine boot-mode=sd -m 1024 -display none -serial null -serial stdio -monitor none \
-device loader,file=u-boot-dtb.bin,addr=0x04000000,cpu-num=0 \
-drive file=zynq-sd.img,format=raw,if=sd -D qemu.log

After applying the fix, success load sd card:

U-Boot 2026.07-rc4-g1e80ee41441c (Jun 15 2026 - 19:05:05 +0800)

Model: Xilinx ZC702 board
DRAM:  ECC disabled 1 GiB
Core:  33 devices, 21 uclasses, devicetree: board
Flash: 0 Bytes
NAND:  0 MiB
MMC:   mmc@e0100000: 0

Loading Environment from FAT... *** Error - No Valid Environment Area found *** Warning - bad env area, using default environment

In:    serial@e0001000
Out:   serial@e0001000
Err:   serial@e0001000

Net:
ZYNQ GEM: e000b000, mdio bus e000b000, phyaddr 7, interface rgmii-id

Warning: ethernet@e000b000 (eth0) using random MAC address - 4a:63:5a:a4:d3:6b
eth0: ethernet@e000b000
Hit any key to stop autoboot: 0
Zynq> fatload mmc 0 ${kernel_addr_r} zImage
10838528 bytes read in 4316 ms (2.4 MiB/s)
Zynq> fatload mmc 0 ${ramdisk_addr_r} rootfs-arm32.cpio.gz
406531 bytes read in 199 ms (1.9 MiB/s)
Zynq> setenv ramdisk_size ${filesize}
Zynq> fatload mmc 0 ${fdt_addr_r} zynq-zc702.dtb
15754 bytes read in 21 ms (732.4 KiB/s)
Zynq> setenv bootargs console=ttyPS0,115200 earlycon ignore_loglevel rdinit=/init Zynq> bootz ${kernel_addr_r} ${ramdisk_addr_r}:${ramdisk_size} ${fdt_addr_r} Kernel image @ 0x2000000 [ 0x000000 - 0xa56200 ]
## Flattened Device Tree blob at 01f00000
Booting using the fdt blob at 0x1f00000
Working FDT set to 1f00000
Loading Ramdisk to 2ff9c000, end 2ffff403 ... OK
Loading Device Tree to 2ff95000, end 2ff9bd89 ... OK
Working FDT set to 2ff95000

Starting kernel ...

---
 hw/sd/sdhci.c | 2 ++
 1 file changed, 2 insertions(+)
hw/sd/sdhci.c +2 -0
index c86dfa281f..0d87289ca4 100644
@@ -644,6 +644,7 @@ static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)
                 s->data_count = 0;
             }
             if (page_aligned && boundary_count == 0) {
+                s->prnsts &= ~SDHC_DOING_READ;
                 break;
             }
         }
@@ -669,6 +670,7 @@ static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)
                 }
             }
             if (page_aligned && boundary_count == 0) {
+                s->prnsts &= ~SDHC_DOING_WRITE;
                 break;
             }
         }
Signature · 1 line
2.43.0
#2 lore raw

Re: [PATCH 1/1] fixed sdma boundary bug

Quoted 25 lines · Tao Ding

During SDMA transfers, the controller raises an interrupt at buffer boundaries to request a system address update. But the emulation fails to clear the SDHC_DOING_READ/WRITE flags after DMA completion, so writes to SDHC_SYSAD are ignored. Clear these flags when the transfer ends.
(according to PartA2_SD_Host_Controller_Simplified_Specification_Ver2.00.pdf section 2.2.1)

This is a QTest in npcm7xx_sdhci-test that exercises this boundary condition and validates the fix.

QTest steps:
1. Replace npcm7xx_sdhci-test.c with the updated version (full code available at: https://pastebin.com/AWVp2wFi).
2. Build with:

       ../configure --target-list="arm-softmmu,aarch64-softmmu" \
                    --disable-slirp --enable-debug --disable-werror
       ninja qemu-system-arm tests/qtest/npcm7xx_sdhci-test
  3. Run the specific test case:
       QTEST_QEMU_BINARY=./qemu-system-arm \
         ./tests/qtest/npcm7xx_sdhci-test \
         -p /arm/npcm7xx_sdhci/read_sd_sdma_boundary_continue

Before the fix, the test fails with a BLKCNT assertion:
ERROR:...wait_for_block_count: assertion failed (qtest_readw(...) == expected): (1 == 0)

Quoted 8 lines

After applying the fix, the test passes:
ok 1 /arm/npcm7xx_sdhci/read_sd_sdma_boundary_continue

The new test case ensures the SDMA boundary handling works correctly and prevents future regressions.

Uboot test:

U-Boot

Quoted 1 line

1. Prepare zImage, uboot and rootfs.cpio.gz, make sure uboot config (MMC_SDHCI_SDMA=y).

Please specify which U-Boot defconfig you used for testing and how zynq-sd.img was generated.

Quoted 66 lines

2. ./qemu-system-aarch64 -M xilinx-zynq-a9 -machine boot-mode=sd -m 1024 -display none -serial null -serial stdio -monitor none \
-device loader,file=u-boot-dtb.bin,addr=0x04000000,cpu-num=0 \
-drive file=zynq-sd.img,format=raw,if=sd -D qemu.log

After applying the fix, success load sd card:

U-Boot 2026.07-rc4-g1e80ee41441c (Jun 15 2026 - 19:05:05 +0800)

Model: Xilinx ZC702 board
DRAM:  ECC disabled 1 GiB
Core:  33 devices, 21 uclasses, devicetree: board
Flash: 0 Bytes
NAND:  0 MiB
MMC:   mmc@e0100000: 0

Loading Environment from FAT... *** Error - No Valid Environment Area found *** Warning - bad env area, using default environment

In:    serial@e0001000
Out:   serial@e0001000
Err:   serial@e0001000

Net:
ZYNQ GEM: e000b000, mdio bus e000b000, phyaddr 7, interface rgmii-id

Warning: ethernet@e000b000 (eth0) using random MAC address - 4a:63:5a:a4:d3:6b
eth0: ethernet@e000b000
Hit any key to stop autoboot: 0
Zynq> fatload mmc 0 ${kernel_addr_r} zImage
10838528 bytes read in 4316 ms (2.4 MiB/s)
Zynq> fatload mmc 0 ${ramdisk_addr_r} rootfs-arm32.cpio.gz
406531 bytes read in 199 ms (1.9 MiB/s)
Zynq> setenv ramdisk_size ${filesize}
Zynq> fatload mmc 0 ${fdt_addr_r} zynq-zc702.dtb
15754 bytes read in 21 ms (732.4 KiB/s)
Zynq> setenv bootargs console=ttyPS0,115200 earlycon ignore_loglevel rdinit=/init Zynq> bootz ${kernel_addr_r} ${ramdisk_addr_r}:${ramdisk_size} ${fdt_addr_r} Kernel image @ 0x2000000 [ 0x000000 - 0xa56200 ]
## Flattened Device Tree blob at 01f00000
Booting using the fdt blob at 0x1f00000
Working FDT set to 1f00000
Loading Ramdisk to 2ff9c000, end 2ffff403 ... OK
Loading Device Tree to 2ff95000, end 2ff9bd89 ... OK
Working FDT set to 2ff95000

Starting kernel ...

---
 hw/sd/sdhci.c | 2 ++
 1 file changed, 2 insertions(+)
hw/sd/sdhci.c +1 -0
index c86dfa281f..0d87289ca4 100644
@@ -644,6 +644,7 @@ static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)
                 s->data_count = 0;
             }
             if (page_aligned && boundary_count == 0) {
+                s->prnsts &= ~SDHC_DOING_READ;
                 break;
             }
         }

hw/sd/sdhci.c static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)

Quoted 8 lines
hw/sd/sdhci.c +1 -1
@@ -669,6 +670,7 @@ static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)
                 }
             }
             if (page_aligned && boundary_count == 0) {
+                s->prnsts &= ~SDHC_DOING_WRITE;
                 break;
             }
         }
--

Regards,
Bin

author
#3 lore raw

Re: [PATCH 1/1] fixed sdma boundary bug

1. u-boot

I used the u-boot obtained fromhttps://github.com/u-boot/u-boot, specifically the commit 1e80ee41441c612f05787a93bbef4e6e422e29d1.

When compiling uboot, I used xilinx_zynq_virt_defconfig and enabled the SDHCI_SDMA function. This can be set through scripts or in manuconfig.

Below is the command I used for local compilation:

$ export ARCH=arm
$ export CROSS_COMPILE=/home/taoding/arm/gcc-arm-11.2-2022.02-x86_64-arm-none-linux-gnueabihf/bin/arm-none-linux-gnueabihf-
$ make O=/home/taoding/arm/u-boot-build-zynq xilinx_zynq_virt_defconfig
$ /home/taoding/arm/u-boot/scripts/config \
    --file /home/taoding/arm/u-boot-build-zynq/.config \
    --enable MMC_SDHCI_SDMA
$ make O=/home/taoding/arm/u-boot-build-zynq olddefconfig
$ make O=/home/taoding/arm/u-boot-build-zynq DEVICE_TREE=zynq-zc702 -j$(nproc)

The U-Boot binary used by QEMU was:
/home/taoding/arm/u-boot-build-zynq/u-boot-dtb.bin

2. linux image
I used Linux 6.6.
Below is the command I used:

$ export ARCH=arm
$ export CROSS_COMPILE=/home/taoding/arm/gcc-arm-11.2-2022.02-x86_64-arm-none-linux-gnueabihf/bin/arm-none-linux-gnueabihf-
$ make multi_v7_defconfig
$ make -j$(nproc) zImage dtbs

The resulting files were:
/home/taoding/arm/linux-6.6/arch/arm/boot/zImage /home/taoding/arm/linux-6.6/arch/arm/boot/dts/xilinx/zynq-zc702.dtb

3. rootfs

I utilized the defconfig and enable static binary to compile and generate rootfs-arm32.cpio.gz

4. zynq-sd.img
The SD image used for testing was a FAT16 image containing:

- zImage
- zynq-zc702.dtb
- rootfs-arm32.cpio.gz

Below is the command I used:

$ truncate -s 128M /home/taoding/arm/zynq-sd.img
$ mkfs.vfat -F 16 /home/taoding/arm/zynq-sd.img
$ mcopy -i /home/taoding/arm/zynq-sd.img \
  /home/taoding/arm/linux-6.6/arch/arm/boot/zImage ::zImage
$ mcopy -i /home/taoding/arm/zynq-sd.img \
  /home/taoding/arm/linux-6.6/arch/arm/boot/dts/xilinx/zynq-zc702.dtb ::zynq-zc702.dtb
$ mcopy -i /home/taoding/arm/zynq-sd.img \
  /home/taoding/arm/rootfs-arm32.cpio.gz ::rootfs-arm32.cpio.gz

If only want to verify the issues fixed in this patch, step 3 can be ignored. After starting qemu and entering uboot, we can manually load a file.
Please note that the driver will only actually use SDMA when the file size is greater than 0.5MB. It is recommended to load zImage from sd instead of dtb.

Thanks,
Tao

Quoted 115 lines · Bin Meng
Quoted 24 lines · Tao Ding

During SDMA transfers, the controller raises an interrupt at buffer boundaries to request a system address update. But the emulation fails to clear the SDHC_DOING_READ/WRITE flags after DMA completion, so writes to SDHC_SYSAD are ignored. Clear these flags when the transfer ends.
(according to PartA2_SD_Host_Controller_Simplified_Specification_Ver2.00.pdf section 2.2.1)

This is a QTest in npcm7xx_sdhci-test that exercises this boundary condition and validates the fix.

QTest steps:
1. Replace npcm7xx_sdhci-test.c with the updated version (full code available at:https://pastebin.com/AWVp2wFi).
2. Build with:

        ../configure --target-list="arm-softmmu,aarch64-softmmu" \
                     --disable-slirp --enable-debug --disable-werror
        ninja qemu-system-arm tests/qtest/npcm7xx_sdhci-test
   3. Run the specific test case:
        QTEST_QEMU_BINARY=./qemu-system-arm \
          ./tests/qtest/npcm7xx_sdhci-test \
          -p /arm/npcm7xx_sdhci/read_sd_sdma_boundary_continue

Before the fix, the test fails with a BLKCNT assertion:

   ERROR:...wait_for_block_count: assertion failed (qtest_readw(...) == expected): (1 == 0)
Quoted 8 lines

After applying the fix, the test passes:
ok 1 /arm/npcm7xx_sdhci/read_sd_sdma_boundary_continue

The new test case ensures the SDMA boundary handling works correctly and prevents future regressions.

Uboot test:

U-Boot

Quoted 1 line

1. Prepare zImage, uboot and rootfs.cpio.gz, make sure uboot config (MMC_SDHCI_SDMA=y).

Please specify which U-Boot defconfig you used for testing and how zynq-sd.img was generated.

Quoted 66 lines

2. ./qemu-system-aarch64 -M xilinx-zynq-a9 -machine boot-mode=sd -m 1024 -display none -serial null -serial stdio -monitor none \
-device loader,file=u-boot-dtb.bin,addr=0x04000000,cpu-num=0 \
-drive file=zynq-sd.img,format=raw,if=sd -D qemu.log

After applying the fix, success load sd card:

U-Boot 2026.07-rc4-g1e80ee41441c (Jun 15 2026 - 19:05:05 +0800)

Model: Xilinx ZC702 board
DRAM:  ECC disabled 1 GiB
Core:  33 devices, 21 uclasses, devicetree: board
Flash: 0 Bytes
NAND:  0 MiB
MMC:   mmc@e0100000: 0

Loading Environment from FAT... *** Error - No Valid Environment Area found *** Warning - bad env area, using default environment

In:    serial@e0001000
Out:   serial@e0001000
Err:   serial@e0001000

Net:
ZYNQ GEM: e000b000, mdio bus e000b000, phyaddr 7, interface rgmii-id

Warning: ethernet@e000b000 (eth0) using random MAC address - 4a:63:5a:a4:d3:6b
eth0: ethernet@e000b000
Hit any key to stop autoboot: 0
Zynq> fatload mmc 0 ${kernel_addr_r} zImage
10838528 bytes read in 4316 ms (2.4 MiB/s)
Zynq> fatload mmc 0 ${ramdisk_addr_r} rootfs-arm32.cpio.gz
406531 bytes read in 199 ms (1.9 MiB/s)
Zynq> setenv ramdisk_size ${filesize}
Zynq> fatload mmc 0 ${fdt_addr_r} zynq-zc702.dtb
15754 bytes read in 21 ms (732.4 KiB/s)
Zynq> setenv bootargs console=ttyPS0,115200 earlycon ignore_loglevel rdinit=/init Zynq> bootz ${kernel_addr_r} ${ramdisk_addr_r}:${ramdisk_size} ${fdt_addr_r} Kernel image @ 0x2000000 [ 0x000000 - 0xa56200 ]
## Flattened Device Tree blob at 01f00000

    Booting using the fdt blob at 0x1f00000
Working FDT set to 1f00000
    Loading Ramdisk to 2ff9c000, end 2ffff403 ... OK
    Loading Device Tree to 2ff95000, end 2ff9bd89 ... OK

Working FDT set to 2ff95000

Starting kernel ...

---
  hw/sd/sdhci.c | 2 ++
  1 file changed, 2 insertions(+)
hw/sd/sdhci.c +1 -0
index c86dfa281f..0d87289ca4 100644
@@ -644,6 +644,7 @@ static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)
                  s->data_count = 0;
              }
              if (page_aligned && boundary_count == 0) {
+                s->prnsts &= ~SDHC_DOING_READ;
                  break;
              }
          }

hw/sd/sdhci.c static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)

Quoted 8 lines
hw/sd/sdhci.c +1 -1
@@ -669,6 +670,7 @@ static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)
                  }
              }
              if (page_aligned && boundary_count == 0) {
+                s->prnsts &= ~SDHC_DOING_WRITE;
                  break;
              }
          }
--

Regards,
Bin

#4 lore raw

Re: [PATCH 1/1] fixed sdma boundary bug

In reply to Tao Ding
Quoted 7 lines · Tao Ding

During SDMA transfers, the controller raises an interrupt at buffer boundaries to request a system address update. But the emulation fails to clear the SDHC_DOING_READ/WRITE flags after DMA completion, so writes to SDHC_SYSAD are ignored. Clear these flags when the transfer ends.
(according to PartA2_SD_Host_Controller_Simplified_Specification_Ver2.00.pdf section 2.2.1)

The commit message could end here, and the following can be put below --- so that they don't go into the commit message while it could still provide useful information for people to understand the issue.

Quoted 6 lines

This is a QTest in npcm7xx_sdhci-test that exercises this boundary condition and validates the fix.

QTest steps:
1. Replace npcm7xx_sdhci-test.c with the updated version (full code available at: https://pastebin.com/AWVp2wFi).

Please post a qtest patch instead.

Quoted 87 lines
  2. Build with:
       ../configure --target-list="arm-softmmu,aarch64-softmmu" \
                    --disable-slirp --enable-debug --disable-werror
       ninja qemu-system-arm tests/qtest/npcm7xx_sdhci-test
  3. Run the specific test case:
       QTEST_QEMU_BINARY=./qemu-system-arm \
         ./tests/qtest/npcm7xx_sdhci-test \
         -p /arm/npcm7xx_sdhci/read_sd_sdma_boundary_continue

Before the fix, the test fails with a BLKCNT assertion:
ERROR:...wait_for_block_count: assertion failed (qtest_readw(...) == expected): (1 == 0)

After applying the fix, the test passes:
ok 1 /arm/npcm7xx_sdhci/read_sd_sdma_boundary_continue

The new test case ensures the SDMA boundary handling works correctly and prevents future regressions.

Uboot test:
1. Prepare zImage, uboot and rootfs.cpio.gz, make sure uboot config (MMC_SDHCI_SDMA=y).
2. ./qemu-system-aarch64 -M xilinx-zynq-a9 -machine boot-mode=sd -m 1024 -display none -serial null -serial stdio -monitor none \
-device loader,file=u-boot-dtb.bin,addr=0x04000000,cpu-num=0 \
-drive file=zynq-sd.img,format=raw,if=sd -D qemu.log

After applying the fix, success load sd card:

U-Boot 2026.07-rc4-g1e80ee41441c (Jun 15 2026 - 19:05:05 +0800)

Model: Xilinx ZC702 board
DRAM:  ECC disabled 1 GiB
Core:  33 devices, 21 uclasses, devicetree: board
Flash: 0 Bytes
NAND:  0 MiB
MMC:   mmc@e0100000: 0

Loading Environment from FAT... *** Error - No Valid Environment Area found *** Warning - bad env area, using default environment

In:    serial@e0001000
Out:   serial@e0001000
Err:   serial@e0001000

Net:
ZYNQ GEM: e000b000, mdio bus e000b000, phyaddr 7, interface rgmii-id

Warning: ethernet@e000b000 (eth0) using random MAC address - 4a:63:5a:a4:d3:6b
eth0: ethernet@e000b000
Hit any key to stop autoboot: 0
Zynq> fatload mmc 0 ${kernel_addr_r} zImage
10838528 bytes read in 4316 ms (2.4 MiB/s)
Zynq> fatload mmc 0 ${ramdisk_addr_r} rootfs-arm32.cpio.gz
406531 bytes read in 199 ms (1.9 MiB/s)
Zynq> setenv ramdisk_size ${filesize}
Zynq> fatload mmc 0 ${fdt_addr_r} zynq-zc702.dtb
15754 bytes read in 21 ms (732.4 KiB/s)
Zynq> setenv bootargs console=ttyPS0,115200 earlycon ignore_loglevel rdinit=/init Zynq> bootz ${kernel_addr_r} ${ramdisk_addr_r}:${ramdisk_size} ${fdt_addr_r} Kernel image @ 0x2000000 [ 0x000000 - 0xa56200 ]
## Flattened Device Tree blob at 01f00000
Booting using the fdt blob at 0x1f00000
Working FDT set to 1f00000
Loading Ramdisk to 2ff9c000, end 2ffff403 ... OK
Loading Device Tree to 2ff95000, end 2ff9bd89 ... OK
Working FDT set to 2ff95000

Starting kernel ...

---
 hw/sd/sdhci.c | 2 ++
 1 file changed, 2 insertions(+)
hw/sd/sdhci.c +1 -0
index c86dfa281f..0d87289ca4 100644
@@ -644,6 +644,7 @@ static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)
                 s->data_count = 0;
             }
             if (page_aligned && boundary_count == 0) {
+                s->prnsts &= ~SDHC_DOING_READ;
                 break;
             }
         }

hw/sd/sdhci.c static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)

Quoted 8 lines
hw/sd/sdhci.c +1 -1
@@ -669,6 +670,7 @@ static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)
                 }
             }
             if (page_aligned && boundary_count == 0) {
+                s->prnsts &= ~SDHC_DOING_WRITE;
                 break;
             }
         }
--

Overall this patch LGTM.

Regards,
Bin

author
#5 lore raw

Re: [PATCH 1/1] fixed sdma boundary bug

Thank you for suggestion.
If post a patch in qtest, I have some doubts about where to place the code.

The bug fixed by this patch is related to generic sdhci. Currently,
qos-test includes a file named sdhci-test.c, which seems to be more appropriately placed here. qos-test conducts tests on many machines, but the way some machines load SD cards varies greatly. Is it possible to target only a specific board, such as xilinx-zynq-a9, and skip the other machines. Another way is to create a new file named xilinx_zynq_sdhci-test.c (similar to npcm7xx_sdhci-test.c).

Thanks,
Tao

Quoted 118 lines · Bin Meng
Quoted 6 lines · Tao Ding

During SDMA transfers, the controller raises an interrupt at buffer boundaries to request a system address update. But the emulation fails to clear the SDHC_DOING_READ/WRITE flags after DMA completion, so writes to SDHC_SYSAD are ignored. Clear these flags when the transfer ends.
(according to PartA2_SD_Host_Controller_Simplified_Specification_Ver2.00.pdf section 2.2.1)

The commit message could end here, and the following can be put below --- so that they don't go into the commit message while it could still provide useful information for people to understand the issue.

Quoted 6 lines

This is a QTest in npcm7xx_sdhci-test that exercises this boundary condition and validates the fix.

QTest steps:
1. Replace npcm7xx_sdhci-test.c with the updated version (full code available at:https://pastebin.com/AWVp2wFi).

Please post a qtest patch instead.

Quoted 87 lines
   2. Build with:
        ../configure --target-list="arm-softmmu,aarch64-softmmu" \
                     --disable-slirp --enable-debug --disable-werror
        ninja qemu-system-arm tests/qtest/npcm7xx_sdhci-test
   3. Run the specific test case:
        QTEST_QEMU_BINARY=./qemu-system-arm \
          ./tests/qtest/npcm7xx_sdhci-test \
          -p /arm/npcm7xx_sdhci/read_sd_sdma_boundary_continue

Before the fix, the test fails with a BLKCNT assertion:

   ERROR:...wait_for_block_count: assertion failed (qtest_readw(...) == expected): (1 == 0)

After applying the fix, the test passes:
ok 1 /arm/npcm7xx_sdhci/read_sd_sdma_boundary_continue

The new test case ensures the SDMA boundary handling works correctly and prevents future regressions.

Uboot test:
1. Prepare zImage, uboot and rootfs.cpio.gz, make sure uboot config (MMC_SDHCI_SDMA=y).
2. ./qemu-system-aarch64 -M xilinx-zynq-a9 -machine boot-mode=sd -m 1024 -display none -serial null -serial stdio -monitor none \
-device loader,file=u-boot-dtb.bin,addr=0x04000000,cpu-num=0 \
-drive file=zynq-sd.img,format=raw,if=sd -D qemu.log

After applying the fix, success load sd card:

U-Boot 2026.07-rc4-g1e80ee41441c (Jun 15 2026 - 19:05:05 +0800)

Model: Xilinx ZC702 board
DRAM:  ECC disabled 1 GiB
Core:  33 devices, 21 uclasses, devicetree: board
Flash: 0 Bytes
NAND:  0 MiB
MMC:   mmc@e0100000: 0

Loading Environment from FAT... *** Error - No Valid Environment Area found *** Warning - bad env area, using default environment

In:    serial@e0001000
Out:   serial@e0001000
Err:   serial@e0001000

Net:
ZYNQ GEM: e000b000, mdio bus e000b000, phyaddr 7, interface rgmii-id

Warning: ethernet@e000b000 (eth0) using random MAC address - 4a:63:5a:a4:d3:6b
eth0: ethernet@e000b000
Hit any key to stop autoboot: 0
Zynq> fatload mmc 0 ${kernel_addr_r} zImage
10838528 bytes read in 4316 ms (2.4 MiB/s)
Zynq> fatload mmc 0 ${ramdisk_addr_r} rootfs-arm32.cpio.gz
406531 bytes read in 199 ms (1.9 MiB/s)
Zynq> setenv ramdisk_size ${filesize}
Zynq> fatload mmc 0 ${fdt_addr_r} zynq-zc702.dtb
15754 bytes read in 21 ms (732.4 KiB/s)
Zynq> setenv bootargs console=ttyPS0,115200 earlycon ignore_loglevel rdinit=/init Zynq> bootz ${kernel_addr_r} ${ramdisk_addr_r}:${ramdisk_size} ${fdt_addr_r} Kernel image @ 0x2000000 [ 0x000000 - 0xa56200 ]
## Flattened Device Tree blob at 01f00000

    Booting using the fdt blob at 0x1f00000
Working FDT set to 1f00000
    Loading Ramdisk to 2ff9c000, end 2ffff403 ... OK
    Loading Device Tree to 2ff95000, end 2ff9bd89 ... OK

Working FDT set to 2ff95000

Starting kernel ...

---
  hw/sd/sdhci.c | 2 ++
  1 file changed, 2 insertions(+)
hw/sd/sdhci.c +1 -0
index c86dfa281f..0d87289ca4 100644
@@ -644,6 +644,7 @@ static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)
                  s->data_count = 0;
              }
              if (page_aligned && boundary_count == 0) {
+                s->prnsts &= ~SDHC_DOING_READ;
                  break;
              }
          }

hw/sd/sdhci.c static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)

Quoted 8 lines
hw/sd/sdhci.c +1 -1
@@ -669,6 +670,7 @@ static void sdhci_sdma_transfer_multi_blocks(SDHCIState *s)
                  }
              }
              if (page_aligned && boundary_count == 0) {
+                s->prnsts &= ~SDHC_DOING_WRITE;
                  break;
              }
          }
--

Overall this patch LGTM.

Regards,
Bin

author
#6 lore raw
In reply to Tao Ding

ping

Quoted 16 lines

Hi all,

Booting Linux from SD card on Xilinx Zynq with SDMA (MMC_SDHCI_SDMA=y) currently hangs during image load.

This patch fixes the hang and enables successful image loading.

Thanks!

Tao Ding (1):
fixed sdma boundary bug

  hw/sd/sdhci.c | 2 ++
  1 file changed, 2 insertions(+)

base-commit: 608ad9307d5b4797d2254ee3a4bad2bc8c8bcbae