txy-21 opened a new pull request, #21950:
URL: https://github.com/apache/nuttx/pull/21950
## Summary
* Why: zipfs is read-only and reads only zip files, with minizip. This PR
reads archives with libarchive (zip by default, other formats by option)
and adds wzipfs, a write-only file system that streams a zip archive
into a
file, a FIFO, a socket or an open file descriptor.
* What: fs/zipfs is rewritten on top of libarchive, wzipfs is new, and the
sim:zipfs configuration and the zipfs document follow.
* How:
- zipfs reads the archive through libarchive callbacks on file_read() and
file_seek(). Opening a file, and seeking backwards, start a new
libarchive reader and scan the archive for the entry. Opening a
directory starts a reader at the beginning of the archive, and
readdir()
returns every entry of the archive, whichever directory is opened.
- wzipfs: open() writes the header of an entry, write() streams the data,
mkdir() writes a directory entry, and unmount writes the central
directory. The target is src=<path> or fd=<N>, and method=store|deflate
selects the compression.
* The 11 commits keep their original authors. They come from a downstream
tree: archivefs is added, extended and renamed to zipfs, and then wzipfs
is
added. Commit 3 changes code that commit 5 replaces.
* libarchive is not in this repository. fs/zipfs/libarchive is a symbolic
link to ../../../apps/system/libarchive, which
https://github.com/apache/nuttx-apps/pull/4076 adds.
* Related: apache/nuttx#17882 and apache/nuttx-apps#3318 were closed. They
proposed the same libarchive approach under the name archivefs.
* BREAKING CHANGE: FS_ZIPFS now depends on UTILS_LIBARCHIVE,
ZIPFS_SEEK_BUFSIZE
is removed, and the fs/zipfs/zlib link is replaced by
fs/zipfs/libarchive.
Quick fix: enable UTILS_LIBARCHIVE (nuttx-apps), and set
FS_ZIPFS_BUFFER_SIZE, the size of the read buffer of each open file.
Depends-On: https://github.com/apache/nuttx-apps/pull/4076
## Impact
* Is new feature added? Is existing feature changed? YES. wzipfs is new.
The
zipfs reader is a different implementation, and the formats are selected
with FS_ZIPFS_FORMAT_*.
* Impact on user (will user need to adapt to change)? YES. See the BREAKING
CHANGE in the Summary. The file mode now comes from the archive:
-rw-rw-r-- for the test archive below, which Linux zip made. Before,
every
file was -r-xr-xr-x, so the files no longer have the execute bit that
"fs/zipfs: Report execute permission for files and the root" added. For
an
archive without a Unix mode, libarchive uses 0664 for files (read from
its
source, not tested).
* Impact on build (will build process change)? YES. FS_ZIPFS needs
system/libarchive of nuttx-apps, which downloads libarchive 3.7.7 at
build
time. The symbolic link is fixed to ../../../apps, so the apps directory
must be named "apps".
* Impact on hardware (will arch(s) / board(s) / driver(s) change)? NO. Only
the sim:zipfs configuration changes.
* Impact on documentation (is update required / provided)? YES. zipfs.rst
is
updated and has a new write mode section. file_permission.rst still lists
zipfs among the file systems that report the execute permission for all
files, which is no longer true. It is not changed here.
* Impact on security (any sort of implications)? YES. libarchive is third
party code that parses archives. Only the zip format is on by default,
and
the option for encrypted entries is off by default.
* Impact on compatibility (backward/forward/interoperability)? YES. See the
BREAKING CHANGE in the Summary. The mount command of zipfs is the same.
* Anything else to consider or add? Each open file and each open directory
of
zipfs allocates a structure with a read buffer of FS_ZIPFS_BUFFER_SIZE
bytes (32768 by default). This was not compared with the memory use of
the
minizip version.
## Testing
I confirm that changes are verified on local setup and works as intended:
* Build Host(s): Linux 7.0.0-34-generic x86_64, Intel i7-14700,
gcc (Ubuntu 13.3.0-6ubuntu2~24.04.1) 13.3.0, GNU Make 4.3, CMake 3.28.3,
Ninja 1.11.1.
* Target(s): sim:zipfs (sim:nsh for the size comparison). Real hardware was
not tested.
* The series and the nuttx-apps pull request were replayed on nuttx
e9eefaa0e9a and nuttx-apps 72937e7f86f. The archive is made on the host
with
"zip -r test.zip a top.txt <a file with a UTF-8 name>" and mounted from
the
host through hostfs.
Testing logs before change (nuttx master), NSH session as printed:
<details><summary>before</summary>
```
nsh> mount -t hostfs -o fs=/tmp/zh.XsVmlD/host /host
nsh> mount -t zipfs -o /host/test.zip /zip
nsh> echo ---ls-zip
---ls-zip
nsh> ls /zip
/zip:
a/
a/1
a/2
top.txt
文件.txt
nsh> echo ---cat-top
---cat-top
nsh> cat /zip/top.txt
top level file
nsh> echo ---cat-a1
---cat-a1
nsh> cat /zip/a/1
this is zipfs test 1
nsh> echo ---cat-utf8
---cat-utf8
nsh> cat /zip/文件.txt
utf8 name file
nsh> echo ---ls-l-top
---ls-l-top
nsh> ls -l /zip/top.txt
-r-xr-xr-x 15 /zip/top.txt
nsh> echo ---ls-l-a1
---ls-l-a1
nsh> ls -l /zip/a/1
-r-xr-xr-x 21 /zip/a/1
nsh> echo ---ls-l-dir
---ls-l-dir
nsh> ls -l /zip/a
nsh: ls: stat failed: 2
nsh> umount /zip
nsh> echo done
done
nsh> poweroff
```
</details>
Testing logs after change (nuttx master and this series), NSH session as
printed:
<details><summary>after</summary>
```
nsh> mount -t hostfs -o fs=/tmp/zh.XsVmlD/host /host
nsh> mount -t zipfs -o /host/test.zip /zip
nsh> echo ---ls-zip
---ls-zip
nsh> ls /zip
/zip:
a/
a/1
a/2
top.txt
文件.txt
nsh> echo ---cat-top
---cat-top
nsh> cat /zip/top.txt
top level file
nsh> echo ---cat-a1
---cat-a1
nsh> cat /zip/a/1
this is zipfs test 1
nsh> echo ---cat-utf8
---cat-utf8
nsh> cat /zip/文件.txt
utf8 name file
nsh> echo ---ls-l-top
---ls-l-top
nsh> ls -l /zip/top.txt
-rw-rw-r-- 15 2026-11-17 12:14 /zip/top.txt
nsh> echo ---ls-l-a1
---ls-l-a1
nsh> ls -l /zip/a/1
-rw-rw-r-- 21 2026-11-17 12:14 /zip/a/1
nsh> echo ---ls-l-dir
---ls-l-dir
nsh> ls -l /zip/a
nsh: ls: stat failed: 2
nsh> umount /zip
nsh> echo done
done
nsh> poweroff
```
</details>
wzipfs: CONFIG_FS_ZIPFS_WRITE=y and CONFIG_FS_ZIPFS_WRITE_DEFLATE=y added
to
sim:zipfs, mounted with the default method=store. NSH session as printed,
and
the check of out.zip on the host:
<details><summary>wzipfs</summary>
```
nsh> mount -t hostfs -o fs=/tmp/zh.XsVmlD/host /host
nsh> mount -t zipfs -o /host/test.zip /zip
nsh> echo ---ls-zip
---ls-zip
nsh> ls /zip
/zip:
a/
a/1
a/2
top.txt
文件.txt
nsh> echo ---cat-top
---cat-top
nsh> cat /zip/top.txt
top level file
nsh> echo ---cat-a1
---cat-a1
nsh> cat /zip/a/1
this is zipfs test 1
nsh> echo ---cat-utf8
---cat-utf8
nsh> cat /zip/文件.txt
utf8 name file
nsh> echo ---ls-l-top
---ls-l-top
nsh> ls -l /zip/top.txt
-rw-rw-r-- 15 2026-11-17 12:14 /zip/top.txt
nsh> echo ---ls-l-a1
---ls-l-a1
nsh> ls -l /zip/a/1
-rw-rw-r-- 21 2026-11-17 12:14 /zip/a/1
nsh> echo ---ls-l-dir
---ls-l-dir
nsh> ls -l /zip/a
nsh: ls: stat failed: 2
nsh> umount /zip
nsh> echo ---wzipfs
---wzipfs
nsh> mount -t wzipfs -o src=/host/out.zip /w
nsh> echo hello > /w/hello.txt
nsh> mkdir /w/dir
nsh> echo nested > /w/dir/n.txt
nsh> umount /w
nsh> echo ---w-done
---w-done
nsh> echo done
done
nsh> poweroff
```
```
Archive: /tmp/zh.XsVmlD/host/out.zip
Length Date Time Name
--------- ---------- ----- ----
6 2008-06-01 08:00 hello.txt
0 2008-06-01 08:00 dir/
7 2008-06-01 08:00 dir/n.txt
--------- -------
13 3 files
No errors detected in compressed data of /tmp/zh.XsVmlD/host/out.zip.
hello.txt => [hello]
dir/n.txt => [nested]
```
</details>
size(1) of nuttx, text/data/bss: sim:zipfs before 671812/2680/9344, after
746168/2644/9376, after with wzipfs 789980/2644/9376. sim:nsh does not
enable
zipfs: 1247954/279696/196800 before (two builds) and after. sim:zipfs is
covered by tools/ci/testlist/sim-03.dat (glob /sim/*/*/*/[o-z]*).
CMake with Ninja builds sim:zipfs with the same series, and its nuttx
binary
mounts and reads the archive the same way.
Not tested: real hardware, ostest, the protected and kernel builds, other
architectures, every commit of the series built on its own, the FIFO,
socket
and fd= targets of wzipfs, an encrypted archive, and DEFLATE at run time.
No regression test is added in this PR. "ls -l" of a directory of the
archive
fails before and after the change.
## PR verification Self-Check
* [ ] This PR introduces only one functional change.
* [x] I have updated all required description fields above.
* [ ] My PR adheres to Contributing
[Guidelines](https://github.com/apache/nuttx/blob/master/CONTRIBUTING.md) and
[Documentation](https://nuttx.apache.org/docs/latest/contributing/index.html)
(git commit title and message, coding standard, etc).
* [x] My PR is still work in progress (not ready for review).
* [ ] My PR is ready for review and can be safely merged into a codebase.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]