drivers/mtd: Optimization of mtdconfig module #18024

zhaoxingyu12 · 2026-01-19T17:59:22Z

Note: Please adhere to Contributing Guidelines.

Summary

Optimize mtdconfig related modules

change filename mtd_config_fs to mtd_config_nvs
change kconfig MTD_CONFIG_FAIL_SAFE to MTD_CONFIG_NVS
other changes related to the name MTD_CONFIG_FAIL_SAFE

Impact

KV read and write functions

Testing

testcases
`
int print_list(void)
{
int fd = open("/dev/nvs", O_RDONLY);

uint8_t buf[255];
struct config_data_s data;
int ret;

data.configdata = buf;
data.len = 255;
ret = ioctl(fd, CFGDIOC_FIRSTCONFIG, &data);
if (ret < 0)
    return ret;

#ifdef CONFIG_MTD_CONFIG_NAMED
syslog(0, "######## key=%s, value=%s", data.name, data.configdata);
#else
syslog(0, "######## id=%d, instance=%d, value=%s", data.id, data.instance, data.configdata);
#endif

while (1) {
    data.configdata = buf;
    data.len = 255;
    ret = ioctl(fd, CFGDIOC_NEXTCONFIG, &data);
    if (ret < 0) {
        ret = -errno;

        /* ENOENT is expected when there are no more entries */

        if (ret == -ENOENT)
            ret = 0;

        break;
    }

#ifdef CONFIG_MTD_CONFIG_NAMED
syslog(0, "######## key=%s, value=%s", data.name, data.configdata);
#else
syslog(0, "######## id=%d, instance=%d, value=%s", data.id, data.instance, data.configdata);
#endif
}

close(fd);
return ret;

}

void test(void)
{
uint64_t filling_id = 0;
struct config_data_s data;
int ret;

int fd = open("/dev/nvs", O_RDWR);
while (1)
  {
  
  uint8_t wr_buf[50] = {0};
  uint8_t wr_buf1[50] = {0};
  snprintf(wr_buf, sizeof(wr_buf), "v----------------------------%llx", filling_id);

#ifdef CONFIG_MTD_CONFIG_NAMED
char rd_buf[10] = {0};

  snprintf(rd_buf, sizeof(rd_buf), "k%llx", filling_id);  
  strlcpy(data.name, rd_buf, sizeof(data.name));

#else
data.id = (uint16_t)filling_id;
data.instance = (int)filling_id + 1;
#endif
data.configdata = wr_buf;
data.len = sizeof(wr_buf);
ret = ioctl(fd, CFGDIOC_SETCONFIG, &data);

   if (ret != 0)
     {
       return;
     }

  data.configdata = wr_buf1;
  ret = ioctl(fd, CFGDIOC_GETCONFIG, &data);

   if (ret < 0)
     {
       assert(0);
     }

   if (memcmp(wr_buf, wr_buf1, sizeof(wr_buf)) != 0)
     {
       assert(0);
     }

   filling_id++;
  }
close(fd);

}

int main(int argc, FAR char *argv[])
{
test();
print_list();
return 0;
}
`

test result
[ 3.724800] [core0] ######## key=k28, value=v----------------------------28
[ 3.725400] [core0] ######## key=k27, value=v----------------------------27
[ 3.726000] [core0] ######## key=k26, value=v----------------------------26
[ 3.726600] [core0] ######## key=k25, value=v----------------------------25
[ 3.727100] [core0] ######## key=k24, value=v----------------------------24
[ 3.727700] [core0] ######## key=k23, value=v----------------------------23
[ 3.728300] [core0] ######## key=k22, value=v----------------------------22
[ 3.728800] [core0] ######## key=k21, value=v----------------------------21
[ 3.729400] [core0] ######## key=k20, value=v----------------------------20
[ 3.729900] [core0] ######## key=k1f, value=v----------------------------1f
[ 3.730500] [core0] ######## key=k1e, value=v----------------------------1e
[ 3.731200] [core0] ######## key=k1d, value=v----------------------------1d
[ 3.731800] [core0] ######## key=k1c, value=v----------------------------1c
[ 3.732500] [core0] ######## key=k1b, value=v----------------------------1b
[ 3.733100] [core0] ######## key=k1a, value=v----------------------------1a
[ 3.733700] [core0] ######## key=k19, value=v----------------------------19
[ 3.734300] [core0] ######## key=k18, value=v----------------------------18
[ 3.734800] [core0] ######## key=k17, value=v----------------------------17
[ 3.735400] [core0] ######## key=k16, value=v----------------------------16
[ 3.736100] [core0] ######## key=k15, value=v----------------------------15
[ 3.736700] [core0] ######## key=k14, value=v----------------------------14
[ 3.737300] [core0] ######## key=k13, value=v----------------------------13
[ 3.737700] [core0] ######## key=k12, value=v----------------------------12
[ 3.738200] [core0] ######## key=k11, value=v----------------------------11
[ 3.738700] [core0] ######## key=k10, value=v----------------------------10
[ 3.739100] [core0] ######## key=kf, value=v----------------------------f
[ 3.739600] [core0] ######## key=ke, value=v----------------------------e
[ 3.740100] [core0] ######## key=kd, value=v----------------------------d
[ 3.740600] [core0] ######## key=kc, value=v----------------------------c
[ 3.741100] [core0] ######## key=kb, value=v----------------------------b
[ 3.741700] [core0] ######## key=ka, value=v----------------------------a
[ 3.742100] [core0] ######## key=k9, value=v----------------------------9
[ 3.742600] [core0] ######## key=k8, value=v----------------------------8
[ 3.743100] [core0] ######## key=k7, value=v----------------------------7
[ 3.743500] [core0] ######## key=k6, value=v----------------------------6
[ 3.744000] [core0] ######## key=k5, value=v----------------------------5
[ 3.744500] [core0] ######## key=k4, value=v----------------------------4
[ 3.744900] [core0] ######## key=k3, value=v----------------------------3
[ 3.745400] [core0] ######## key=k2, value=v----------------------------2
[ 3.746000] [core0] ######## key=k1, value=v----------------------------1
[ 3.747300] [core0] ######## key=k0, value=v----------------------------0
[ 3.748300] [core0] ######## key=k29, value=v----------------------------29
[ 3.748700] [core0] ######## key=k2a, value=v----------------------------2a
[ 3.749300] [core0] ######## key=k2b, value=v----------------------------2b
[ 3.749800] [core0] ######## key=k2c, value=v----------------------------2c
[ 3.750300] [core0] ######## key=k2d, value=v----------------------------2d
[ 3.751000] [core0] ######## key=k2e, value=v----------------------------2e
[ 3.751600] [core0] ######## key=k2f, value=v----------------------------2f
[ 3.752200] [core0] ######## key=k30, value=v----------------------------30
[ 3.752700] [core0] ######## key=k31, value=v----------------------------31
[ 3.753300] [core0] ######## key=k32, value=v----------------------------32
[ 3.753800] [core0] ######## key=k33, value=v----------------------------33
[ 3.754300] [core0] ######## key=k34, value=v----------------------------34
[ 3.754900] [core0] ######## key=k35, value=v----------------------------35
[ 3.755400] [core0] ######## key=k36, value=v----------------------------36
[ 3.756100] [core0] ######## key=k37, value=v----------------------------37
[ 3.756600] [core0] ######## key=k38, value=v----------------------------38
[ 3.757200] [core0] ######## key=k39, value=v----------------------------39
[ 3.757700] [core0] ######## key=k3a, value=v----------------------------3a
[ 3.758200] [core0] ######## key=k3b, value=v----------------------------3b
[ 3.758800] [core0] ######## key=k3c, value=v----------------------------3c
[ 3.759300] [core0] ######## key=k3d, value=v----------------------------3d
[ 3.759800] [core0] ######## key=k3e, value=v----------------------------3e
[ 3.760300] [core0] ######## key=k3f, value=v----------------------------3f
[ 3.761100] [core0] ######## key=k40, value=v----------------------------40
[ 3.761700] [core0] ######## key=k41, value=v----------------------------41
[ 3.762200] [core0] ######## key=k42, value=v----------------------------42
[ 3.762700] [core0] ######## key=k43, value=v----------------------------43
[ 3.763200] [core0] ######## key=k44, value=v----------------------------44
[ 3.763800] [core0] ######## key=k45, value=v----------------------------45
[ 3.764300] [core0] ######## key=k46, value=v----------------------------46
[ 3.764800] [core0] ######## key=k47, value=v----------------------------47
[ 3.765400] [core0] ######## key=k48, value=v----------------------------48
[ 3.766000] [core0] ######## key=k49, value=v----------------------------49
[ 3.766600] [core0] ######## key=k4a, value=v----------------------------4a
[ 3.767100] [core0] ######## key=k4b, value=v----------------------------4b
[ 3.767600] [core0] ######## key=k4c, value=v----------------------------4c
[ 3.768200] [core0] ######## key=k4d, value=v----------------------------4d
[ 3.768800] [core0] ######## key=k4e, value=v----------------------------4e
[ 3.769300] [core0] ######## key=k4f, value=v----------------------------4f
[ 3.769900] [core0] ######## key=k50, value=v----------------------------50
[ 3.771100] [core0] ######## key=k51, value=v----------------------------51

Laczen · 2026-01-19T20:26:06Z

@zhaoxingyu12 does NuttX have the possibility to distinguish between devices that require erase and devices that don't? Even for devices that do not require erase the benefits of zms over nvs (emulating a erase by writing 0xff) is very limited. I am unsure that the addition of zms (with associated maintenance) can be justified.

xiaoxiang781216 · 2026-01-20T06:20:55Z

@zhaoxingyu12 does NuttX have the possibility to distinguish between devices that require erase and devices that don't? Even for devices that do not require erase the benefits of zms over nvs (emulating a erase by writing 0xff) is very limited. I am unsure that the addition of zms (with associated maintenance) can be justified.

@Laczen you can check whether mtd driver implement mtd_dev_s::erase.

Laczen · 2026-01-20T07:50:53Z

@zhaoxingyu12 does NuttX have the possibility to distinguish between devices that require erase and devices that don't? Even for devices that do not require erase the benefits of zms over nvs (emulating a erase by writing 0xff) is very limited. I am unsure that the addition of zms (with associated maintenance) can be justified.

@Laczen you can check whether mtd driver implement mtd_dev_s::erase.

@xiaoxiang781216 this could be done, but this might break other drivers. E.g. the rwbuffer directly calls the underlying mtd_erase method.

xiaoxiang781216 · 2026-01-20T09:13:29Z

@zhaoxingyu12 does NuttX have the possibility to distinguish between devices that require erase and devices that don't? Even for devices that do not require erase the benefits of zms over nvs (emulating a erase by writing 0xff) is very limited. I am unsure that the addition of zms (with associated maintenance) can be justified.

@Laczen you can check whether mtd driver implement mtd_dev_s::erase.

@xiaoxiang781216 this could be done, but this might break other drivers. E.g. the rwbuffer directly calls the underlying mtd_erase method.

All places already add the check, I think. @zhaoxingyu12 please confirm this.

improve the speed of finding the target ate Signed-off-by: guohao15 <guohao15@xiaomi.com>

change filename mtd/mtd_config_fs.c to mtd/mtd_config_nvs.c and optimize the configuration related to mtdconfig Signed-off-by: zhaoxingyu1 <zhaoxingyu1@xiaomi.com>

For nvs test in qemu Signed-off-by: guohao15 <guohao15@xiaomi.com>

The structure of mtd_config in Kconfig has changed. Signed-off-by: zhaoxingyu1 <zhaoxingyu1@xiaomi.com>

…y_path api mtd_config/mtd_config_nvs should all support mtdconfig_register_by_path/mtdconfig_unregister_by_path api Signed-off-by: zhaoxingyu1 <zhaoxingyu1@xiaomi.com>

For the MTD_CONFIG configuration item in drivers/mtd/Kconfig has changed. Signed-off-by: zhaoxingyu1 <zhaoxingyu1@xiaomi.com>

xiaoxiang781216 · 2026-01-20T14:31:47Z

@zhaoxingyu12 please fix the following error:

====================================================================================
Configuration/Tool: esp32c3-legacy-devkit/nvcfgdata
2026-01-20 13:56:42
------------------------------------------------------------------------------------
  Cleaning...
  Configuring...
  Building NuttX...
  [1/1] Normalize esp32c3-legacy-devkit/nvcfgdata
46d45
< CONFIG_SYSTEM_CFGDATA=y
48d46
< CONFIG_TESTING_MTD_CONFIG_FAIL_SAFE=y
Saving the new configuration file
HEAD detached at pull/18024/merge
Changes not staged for commit:
  (use "git add <file>..." to update what will be committed)
  (use "git restore <file>..." to discard changes in working directory)
	modified:   boards/risc-v/esp32c3-legacy/esp32c3-legacy-devkit/configs/nvcfgdata/defconfig

michallenc · 2026-01-20T15:43:49Z

Shouldn't this be marked as breaking because of MTD_CONFIG_FAIL_SAFE removal?

zhaoxingyu12 · 2026-01-20T15:51:04Z

@zhaoxingyu12 please fix the following error:

====================================================================================
Configuration/Tool: esp32c3-legacy-devkit/nvcfgdata
2026-01-20 13:56:42
------------------------------------------------------------------------------------
  Cleaning...
  Configuring...
  Building NuttX...
  [1/1] Normalize esp32c3-legacy-devkit/nvcfgdata
46d45
< CONFIG_SYSTEM_CFGDATA=y
48d46
< CONFIG_TESTING_MTD_CONFIG_FAIL_SAFE=y
Saving the new configuration file
HEAD detached at pull/18024/merge
Changes not staged for commit:
  (use "git add <file>..." to update what will be committed)
  (use "git restore <file>..." to discard changes in working directory)
	modified:   boards/risc-v/esp32c3-legacy/esp32c3-legacy-devkit/configs/nvcfgdata/defconfig

Does it mean to remove these two configs in boards/risc-v/esp32c3-legacy/esp32c3-legacy-devkit/configs/nvcfgdata/defconfig?
CONFIG_SYSTEM_CFGDATA=y
CONFIG_TESTING_MTD_CONFIG_FAIL_SAFE=y

i think #18024 and apache/nuttx-apps#3344 compiling together can fix it . but i dont know how to put the nuttx and nuttx-apps cict together

zhaoxingyu12 · 2026-01-20T16:34:17Z

@zhaoxingyu12 does NuttX have the possibility to distinguish between devices that require erase and devices that don't? Even for devices that do not require erase the benefits of zms over nvs (emulating a erase by writing 0xff) is very limited. I am unsure that the addition of zms (with associated maintenance) can be justified.

@Laczen you can check whether mtd driver implement mtd_dev_s::erase.

@xiaoxiang781216 this could be done, but this might break other drivers. E.g. the rwbuffer directly calls the underlying mtd_erase method.

All places already add the check, I think. @zhaoxingyu12 please confirm this.

@xiaoxiang781216 @Laczen It is currently only used in mtd_partition, and has (part->child.erase = mtd->erase ? part_erase : NULL; ) check in mtd_partition.
I will fully check other drivers.

xiaoxiang781216 · 2026-01-21T02:28:19Z

@zhaoxingyu12 please fix:

====================================================================================
Configuration/Tool: esp32c3-legacy-devkit/nvcfgdata
2026-01-20 13:56:42
------------------------------------------------------------------------------------
  Cleaning...
  Configuring...
  Building NuttX...
  [1/1] Normalize esp32c3-legacy-devkit/nvcfgdata
46d45
< CONFIG_SYSTEM_CFGDATA=y
48d46
< CONFIG_TESTING_MTD_CONFIG_FAIL_SAFE=y

zhaoxingyu12 requested review from btashton, hartmannathan, jerpelea, michallenc, pkarashchenko, raiden00pl, simbit18 and xiaoxiang781216 as code owners January 19, 2026 17:59

github-actions bot added Area: Drivers Drivers issues Area: File System File System issues Board: arm Board: risc-v Size: XL The size of the change in this PR is very large. Consider breaking down the PR into smaller pieces. labels Jan 19, 2026

zhaoxingyu12 force-pushed the master branch from 4ee4465 to 56a1d8e Compare January 19, 2026 18:23

zhaoxingyu12 force-pushed the master branch from 56a1d8e to 23e8ff3 Compare January 20, 2026 02:12

zhaoxingyu12 closed this Jan 20, 2026

zhaoxingyu12 reopened this Jan 20, 2026

zhaoxingyu12 force-pushed the master branch from 23e8ff3 to 6e9cfe9 Compare January 20, 2026 03:07

zhaoxingyu12 changed the title ~~add zms module for RRAM storage device~~ Optimization of mtdconfig module Jan 20, 2026

zhaoxingyu12 force-pushed the master branch from 6e9cfe9 to 2b0fa5f Compare January 20, 2026 03:13

github-actions bot added the Size: M The size of the change in this PR is medium label Jan 20, 2026

xiaoxiang781216 approved these changes Jan 20, 2026

View reviewed changes

jerpelea changed the title ~~Optimization of mtdconfig module~~ drivers/mtd: Optimization of mtdconfig module Jan 20, 2026

jerpelea approved these changes Jan 20, 2026

View reviewed changes

guohao15 and others added 2 commits January 20, 2026 17:47

mtd/nvs: add nvs cache buffer

cfcf4f0

improve the speed of finding the target ate Signed-off-by: guohao15 <guohao15@xiaomi.com>

mtdconfig: modify mtdconfig kconfig

3f4e8e1

change filename mtd/mtd_config_fs.c to mtd/mtd_config_nvs.c and optimize the configuration related to mtdconfig Signed-off-by: zhaoxingyu1 <zhaoxingyu1@xiaomi.com>

guohao15 and others added 3 commits January 20, 2026 17:47

mtdconfig: support ram_mtdconfig device && lomtdconfig device

f947881

For nvs test in qemu Signed-off-by: guohao15 <guohao15@xiaomi.com>

mtdconfig: lomtdconfig device change to depends on !MTD_CONFIG_NONE

cf46299

The structure of mtd_config in Kconfig has changed. Signed-off-by: zhaoxingyu1 <zhaoxingyu1@xiaomi.com>

mtd/mtd_config: add mtdconfig_register_by_path/mtdconfig_unregister_b…

797ddef

…y_path api mtd_config/mtd_config_nvs should all support mtdconfig_register_by_path/mtdconfig_unregister_by_path api Signed-off-by: zhaoxingyu1 <zhaoxingyu1@xiaomi.com>

zhaoxingyu12 force-pushed the master branch from 2b0fa5f to f7924e3 Compare January 20, 2026 09:47

mtdconfig: change the relevant judgments of mtd_config

2fc3af9

For the MTD_CONFIG configuration item in drivers/mtd/Kconfig has changed. Signed-off-by: zhaoxingyu1 <zhaoxingyu1@xiaomi.com>

zhaoxingyu12 force-pushed the master branch from f7924e3 to 2fc3af9 Compare January 20, 2026 13:12

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

drivers/mtd: Optimization of mtdconfig module #18024

drivers/mtd: Optimization of mtdconfig module #18024

zhaoxingyu12 commented Jan 19, 2026 •

edited

Loading

Uh oh!

Laczen commented Jan 19, 2026

Uh oh!

xiaoxiang781216 commented Jan 20, 2026

Uh oh!

Laczen commented Jan 20, 2026

Uh oh!

xiaoxiang781216 commented Jan 20, 2026

Uh oh!

xiaoxiang781216 commented Jan 20, 2026

Uh oh!

michallenc commented Jan 20, 2026

Uh oh!

zhaoxingyu12 commented Jan 20, 2026 •

edited

Loading

Uh oh!

zhaoxingyu12 commented Jan 20, 2026

Uh oh!

xiaoxiang781216 commented Jan 21, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

drivers/mtd: Optimization of mtdconfig module #18024

Are you sure you want to change the base?

drivers/mtd: Optimization of mtdconfig module #18024

Conversation

zhaoxingyu12 commented Jan 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Impact

Testing

Uh oh!

Laczen commented Jan 19, 2026

Uh oh!

xiaoxiang781216 commented Jan 20, 2026

Uh oh!

Laczen commented Jan 20, 2026

Uh oh!

xiaoxiang781216 commented Jan 20, 2026

Uh oh!

xiaoxiang781216 commented Jan 20, 2026

Uh oh!

michallenc commented Jan 20, 2026

Uh oh!

zhaoxingyu12 commented Jan 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

zhaoxingyu12 commented Jan 20, 2026

Uh oh!

xiaoxiang781216 commented Jan 21, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

zhaoxingyu12 commented Jan 19, 2026 •

edited

Loading

zhaoxingyu12 commented Jan 20, 2026 •

edited

Loading