Skip to content

fs/inode: add POSIX path-resolution infrastructure - #19950

Merged
GUIDINGLI merged 9 commits into
apache:masterfrom
Kaben123:pse52-02-path-resolve
Aug 28, 2026
Merged

fs/inode: add POSIX path-resolution infrastructure#19950
GUIDINGLI merged 9 commits into
apache:masterfrom
Kaben123:pse52-02-path-resolve

Conversation

@Kaben123

@Kaben123 Kaben123 commented Aug 24, 2026

Copy link
Copy Markdown

Summary

This PR adds POSIX path-resolution infrastructure to inode_search(). It is
independent of the hardlink infrastructure PR (both are based on master and
can be reviewed in parallel), and provides the path-canonicalization
primitives that later bug-fix PRs depend on.

Changes (one upstream commit each), all touching only
fs/inode/fs_inodesearch.c and fs/inode/inode.h:

  1. fs/inode: format link path before resolving link target
  2. fs/inode: return ENOENT if pathname is empty in inode_search
  3. fs/inode: return ENOTDIR if a path prefix is not a directory
  4. fs/inode: change fs_heap to lib_get_tempbuffer/lib_put_tempbuffer —
    moves the inode search-path buffer allocation onto the temp-buffer pool
    (updates the RELEASE_SEARCH macro and desc->buffer). This keeps the
    allocator consistent across the whole path-resolution logic.
  5. fs/inode: support relative path when inode_search
  6. fs/inode: support path ending with '..' and '.' in inode_search
  7. fs/inode: return ENAMETOOLONG for path/filename longer than NAME_MAX
  8. fs/inode: fix off-by-one error in _inode_checkpath NAME_MAX check

Impact

  • Path resolution now returns the correct POSIX error codes
    (ENOENT/ENOTDIR/ENAMETOOLONG) and correctly handles relative paths
    and ./.. segments.
  • The search-path buffer now uses lib_get_tempbuffer/lib_put_tempbuffer
    consistently (commit 4), so the allocation and release paths match.
  • No public API change; contained entirely within the inode search layer.

Testing

  • Built sim:nsh — zero errors/warnings.
  • tools/checkpatch.sh passes on all changed lines.

This is part of a filesystem enhancement stack. It is independent of the
hardlink infrastructure PR and can be merged in either order. A later bug-fix
PR depends on both this PR and the hardlink/backends PRs.

@github-actions github-actions Bot added Area: File System File System issues Size: M The size of the change in this PR is medium labels Aug 24, 2026
@github-actions

github-actions Bot commented Aug 24, 2026 •
edited
Loading

Copy link
Copy Markdown

MemBrowse Memory Report

arduino-mega2560

  • flash: .text +490 B (+0.8%, 64,860 B / 262,144 B, total: 25% used)

esp32-devkitc

  • ROM: .flash.text +236 B (+0.2%, 124,668 B / 4,194,272 B, total: 3% used)
  • irom0_0_seg: .flash.text +236 B (+0.3%, 88,872 B / 3,342,304 B, total: 3% used)

hifive1-revb

  • flash: .text +212 B (+0.3%, 83,584 B / 4,194,304 B, total: 2% used)

mirtoo

  • kseg0_progmem: .text +564 B (+0.8%, 67,532 B / 131,072 B, total: 52% used)

qemu-armv8a

  • Code: .text.inode_search +216 B, .text.rename +8 B (+0.1%, 319,404 B)

qemu-intel64

  • Code: .text +430 B (+0.0%, 8,659,494 B)

rx65n-rsk2mb

  • ROM: .text +224 B (+0.3%, 86,688 B / 2,097,152 B, total: 4% used)

s698pm-dkit

  • Code: .text +912 B (+0.3%, 364,464 B)

stm32-nucleo-f103rb

  • flash: .text +176 B (+0.5%, 34,276 B / 131,072 B, total: 26% used)

@xiaoxiang781216

xiaoxiang781216 commented Aug 24, 2026

Copy link
Copy Markdown
Contributor

@Kaben123 please fix:

../nuttx/tools/checkpatch.sh -c -u -m -g  fae893d4115e43fcf530243b391f7708cc71a6af..HEAD
❌ Missing git commit message
❌ Missing git commit message
❌ Missing git commit message
❌ Missing git commit message
❌ Missing git commit message
❌ Missing git commit message

@Kaben123
Kaben123 force-pushed the pse52-02-path-resolve branch from 066e6d8 to 661c2d0 Compare August 25, 2026 02:27
@apache apache deleted a comment from Kaben123 Aug 26, 2026
@apache apache deleted a comment from Kaben123 Aug 26, 2026
@Kaben123
Kaben123 force-pushed the pse52-02-path-resolve branch from 0d54467 to 18127da Compare August 26, 2026 06:36
@Kaben123

Kaben123 commented Aug 26, 2026

Copy link
Copy Markdown
Author

@xiaoxiang781216 Could you help re-run CI on this PR when convenient?

The only failing job — Linux (arm-08) — is not a code issue. It fails in the Download Source Artifact / Run builds step with:

digest-mismatch: error
Failed to clone nuttx-ntfc-testing after N attempts

This is a CI infrastructure hiccup cloning the nuttx-ntfc-testing helper repo, not a build/compile/style problem. The log shows no compile error, link error, or assertion — the job simply couldn't fetch the test repo. All other 22 jobs pass.

Note the previously failing esp32s3-devkit:usbmsc (xtensa-03) esptool --ram-only-header alignment issue is now resolved by the defconfig change in this PR. I don't have admin rights to trigger a re-run myself.

@simbit18

simbit18 commented Aug 26, 2026

Copy link
Copy Markdown
Contributor

Hi @Kaben123 please rebase

@Kaben123
Kaben123 force-pushed the pse52-02-path-resolve branch from 18127da to 0ccee60 Compare August 27, 2026 01:42
@Kaben123
Kaben123 requested a review from yamt as a code owner August 27, 2026 03:28
@Kaben123
Kaben123 force-pushed the pse52-02-path-resolve branch from 2f0c845 to 2530ca7 Compare August 27, 2026 06:22
xiaoxiang781216
xiaoxiang781216 reviewed Aug 27, 2026
View reviewed changes
Comment thread boards/sim/sim/sim/configs/lua/defconfig
@Kaben123
Kaben123 force-pushed the pse52-02-path-resolve branch from 2530ca7 to 78b772d Compare August 27, 2026 06:49
@xiaoxiang781216

xiaoxiang781216 commented Aug 27, 2026

Copy link
Copy Markdown
Contributor

ci was fixed, please rebase your change to the last master. @Kaben123

@Zfeather914 yukangzhi
fs/inode: format link path before resolving link target
7008792 
When resolving a symbolic link target, call the public inode_search()
instead of the internal _inode_search() so that the link target path is
first formatted (leading '/' handling and relative-path conversion)
before the lookup.  This ensures link targets are resolved through the
same normalization path as ordinary lookups.

Signed-off-by: zhengyu16 <zhengyu16@xiaomi.com>
Zfeather914 and others added 8 commits August 27, 2026 19:28
@Zfeather914 yukangzhi
fs/inode: return ENOENT if pathname is empty in inode_search
86fd925 
An empty pathname does not name any inode.  Return -ENOENT early in
inode_search() when the path is an empty string, instead of continuing
into the search logic with a zero-length path.

Signed-off-by: zhengyu16 <zhengyu16@xiaomi.com>
@Zfeather914 yukangzhi
fs/inode: return ENOTDIR if a path prefix is not a directory
e932758 
While walking the path components, a non-final component must refer to a
directory.  When descending into a child, verify the parent inode is a
pseudo directory; if it is not, stop the search and return -ENOTDIR as
required by POSIX for a path prefix that is not a directory.

Signed-off-by: zhengyu16 <zhengyu16@xiaomi.com>
@zhaoxingyu12 yukangzhi
fs/inode: change fs_heap to lib_get_tempbuffer/lib_put_tempbuffer
a1eabc9 
Replace the fs_heap_asprintf()/fs_heap_free() based allocation of the
path buffer in the inode search with the lib_get_tempbuffer()/
lib_put_tempbuffer() pool.  Fixed PATH_MAX sized temporary buffers avoid
per-call heap allocation and keep the buffer allocator consistent with
the rest of the path-resolution code.

Signed-off-by: zhaoxingyu1 <zhaoxingyu1@xiaomi.com>
@zhaoxingyu12 yukangzhi
fs/inode: support relative path when inode_search
8f37e90 
Add support for resolving relative path components (in particular the
".." parent references) during the inode search.  A helper
_compute_path_depth() computes the remaining path depth so that a mount
point is only treated as the terminal node when the depth is positive,
and "../" components walk back up to the parent inode.

Signed-off-by: zhaoxingyu1 <zhaoxingyu1@xiaomi.com>
@zhaoxingyu12 yukangzhi
fs/inode: support path ending with '..' and '.' in inode_search
fc0c933 
example: stat(".", buf) and stat("..", buf)

Signed-off-by: zhaoxingyu1 <zhaoxingyu1@xiaomi.com>
@guohao15 yukangzhi
fs/inode: return ENAMETOOLONG for path/filename longer than NAME_MAX
8dd94a7 
Add a helper _inode_checkpath() that validates the path before the
search: it returns -ENOENT for an empty path and -ENAMETOOLONG when any
single path component exceeds NAME_MAX or the whole path exceeds
PATH_MAX.  inode_search() now runs this check first so that oversized
paths and file names are rejected with the correct POSIX error code.

Signed-off-by: guohao15 <guohao15@xiaomi.com>
@WangXX-mi yukangzhi
fs/inode: fix off-by-one error in _inode_checkpath NAME_MAX check
5bdaf4d 
The _inode_checkpath function uses `namelen < NAME_MAX` to validate
path segment lengths. When a filename is exactly NAME_MAX characters
long and is followed by more path segments (e.g. /dir/), the loop
exits with namelen == NAME_MAX before processing the '/' separator,
causing a spurious ENAMETOOLONG error.

Per POSIX, NAME_MAX is the maximum number of bytes in a filename not
including the terminating null, so a filename of exactly NAME_MAX
characters is valid. Change the condition to `namelen <= NAME_MAX`
so the loop can process the trailing '/' separator and correctly
reset namelen for the next path segment.

Signed-off-by: wangxingxing <wangxingxing@xiaomi.com>
yukangzhi
boards/esp32s3-devkit: disable C++ in usbmsc to avoid esptool align a…
c036807 
…ssert

The esp32s3-devkit:usbmsc config uses CONFIG_ESPRESSIF_SIMPLE_BOOT, so
the image is packaged with "esptool elf2image --ram-only-header".  With
esptool v5.2.0, that path asserts that each flash segment satisfies
  (f.tell() + 8 + BOOTLOADER_FLASH_OFFSET) % IROM_ALIGN
    == segment.addr % IROM_ALIGN
When .flash.text lands right on a 64 KB (IROM_ALIGN) boundary the
assertion cannot be satisfied and elf2image fails with AssertionError,
even though NuttX itself links cleanly.

usbmsc is a plain C test and does not need the C++ runtime.  Disabling
CONFIG_HAVE_CXX / CONFIG_HAVE_CXXINITIALIZE moves .flash.text off the
IROM_ALIGN boundary so the SIMPLE_BOOT image can be generated, without
changing the boot mode or the test's intent.

Signed-off-by: yukangzhi <yukangzhi@xiaomi.com>
@Kaben123
Kaben123 force-pushed the pse52-02-path-resolve branch from 78b772d to c036807 Compare August 27, 2026 11:29
@GUIDINGLI
GUIDINGLI merged commit b42ec9a into apache:master Aug 28, 2026
53 checks passed
This was referenced Aug 28, 2026
Merged
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@xiaoxiang781216 xiaoxiang781216 xiaoxiang781216 approved these changes
@GUIDINGLI GUIDINGLI GUIDINGLI approved these changes
@Donny9 Donny9 Awaiting requested review from Donny9 Donny9 is a code owner
@pkarashchenko pkarashchenko Awaiting requested review from pkarashchenko pkarashchenko is a code owner
@jerpelea jerpelea Awaiting requested review from jerpelea jerpelea is a code owner
@acassis acassis Awaiting requested review from acassis
@simbit18 simbit18 Awaiting requested review from simbit18
@yamt yamt Awaiting requested review from yamt

Assignees

No one assigned

Labels

Area: File System File System issues Board: xtensa Size: M The size of the change in this PR is medium

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

8 participants

@Kaben123 @xiaoxiang781216 @simbit18 @GUIDINGLI @Zfeather914 @zhaoxingyu12 @guohao15 @WangXX-mi