Skip to contentSkip to conversation

fastboot: Fix has-slot command always returning yes for fb_nand

PATCH Superseded Chance Yang 2 participants 2 messages Jul 8, 2025 – Jul 11, 2025

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 fastboot: Fix has-slot command always returning yes for fb_nand

The issue was a mismatch in return value conventions between functions:
- getvar_get_part_info() expects >= 0 for success
- fb_nand_lookup() returns 0 on success, 1 on failure (from find_dev_and_part)

When partition didn't exist, fb_nand_lookup returned 1, but fastboot_nand_get_part_info passed it directly to getvar_get_part_info, which treated 1 >= 0 as success, causing has-slot to always return yes.

Fix by converting positive return values to -ENOENT in fastboot_nand_get_part_info to match the expected error convention.

---
 drivers/fastboot/fb_nand.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/fastboot/fb_nand.c +6 -1
index afc64fd5280717ae4041ed70268ccc01cfbb0496..b819541eb8fa6f537ca40fb3ea81bc7aab6118bf 100644
@@ -157,8 +157,13 @@ int fastboot_nand_get_part_info(const char *part_name,
 				struct part_info **part_info, char *response)
 {
 	struct mtd_info *mtd = NULL;
+	int ret;
+
+	ret = fb_nand_lookup(part_name, &mtd, part_info, response);
+	if (ret > 0)
+		return -ENOENT;
 
-	return fb_nand_lookup(part_name, &mtd, part_info, response);
+	return ret;
 }
 
 /**

---
base-commit: d1d53c252a4a746db5ebcdf0d6de3aa0feec504e
change-id: 20250708-master-b6a53395df05

Best regards,

Signature · 1 line
Chance Yang <[email protected]>
#1 lore raw

Hi Chance,

Thank you for the patch.

Quoted 22 lines · Chance Yang

The issue was a mismatch in return value conventions between functions:
- getvar_get_part_info() expects >= 0 for success
- fb_nand_lookup() returns 0 on success, 1 on failure (from find_dev_and_part)

When partition didn't exist, fb_nand_lookup returned 1, but fastboot_nand_get_part_info passed it directly to getvar_get_part_info, which treated 1 >= 0 as success, causing has-slot to always return yes.

Fix by converting positive return values to -ENOENT in fastboot_nand_get_part_info to match the expected error convention.

---
 drivers/fastboot/fb_nand.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/fastboot/fb_nand.c +0 -0
index afc64fd5280717ae4041ed70268ccc01cfbb0496..b819541eb8fa6f537ca40fb3ea81bc7aab6118bf 100644

drivers/fastboot/fb_nand.c int fastboot_nand_get_part_info(const char *part_name, · line 163

Quoted 7 lines
drivers/fastboot/fb_nand.c +4 -0
@@ -157,8 +157,13 @@ int fastboot_nand_get_part_info(const char *part_name,
 				struct part_info **part_info, char *response)
 {
 	struct mtd_info *mtd = NULL;
+	int ret;
+
+	ret = fb_nand_lookup(part_name, &mtd, part_info, response);
+	if (ret > 0)

ret > 0 does not cover all the failure paths in fb_nand_lookup():

Inspecting fb_nand_loopup(), it can fail in 3 different ways. In case of failure:

1. mtdparts_init(): returns 1
2. find_dev_and_part(): returns 1
3. dev->id->type != MTD_DEV_TYPE_NAND: returns -EINVAL

Right now, we only cover failure path 1. and 2.

Could we change this to:

if (ret)

Thanks,
Mattijs

Quoted 15 lines
drivers/fastboot/fb_nand.c +2 -2
+		return -ENOENT;
 
-	return fb_nand_lookup(part_name, &mtd, part_info, response);
+	return ret;
 }
 
 /**
 
---

base-commit: d1d53c252a4a746db5ebcdf0d6de3aa0feec504e
change-id: 20250708-master-b6a53395df05

Best regards,

Signature · 1 line
Chance Yang <[email protected]>