Add file:get_cwd/0 #1806

mat-hek · 2025-08-20T12:40:59Z

These changes are made under both the "Apache 2.0" and the "GNU Lesser General
Public License 2.1 or later" license terms (dual license).

SPDX-License-Identifier: Apache-2.0 OR LGPL-2.1-or-later

pguyot · 2025-08-20T20:06:04Z

AFAIK, prim_file is not part of OTP documented library.
Shouldn't we implement file:get_cwd() instead?

mat-hek · 2025-08-21T10:19:21Z

@pguyot indeed it's not, but we need it because it's called by the OTP internals, and not only by the file server

pguyot · 2025-08-21T17:15:01Z

@pguyot indeed it's not, but we need it because it's called by the OTP internals, and not only by the file server

Which module calls it? I searched OTP source code (for prim_file:get_cwd and ?PRIM_FILE:get_cwd) and didn't find any relevant call.

mat-hek · 2025-08-22T13:54:39Z

@pguyot well, now I can't find it either... changed to file:get_cwd, I'll reiterate if prim_file pops out again

bettio · 2025-09-07T17:24:19Z

src/libAtomVM/nifs.c

+        }
+        term result_tuple = term_alloc_tuple(2, &ctx->heap);
+        term reason = UNDEFINED_ATOM;
+        switch (errno) {


POSIX error code are already standard in OTP, so we don't have to invent new codes:
https://www.erlang.org/doc/apps/kernel/file.html#t:posix/0
Also we have a functions that handles them, that is posix_errno_to_term.

bettio · 2025-09-07T17:40:09Z

src/libAtomVM/nifs.c

+    UNUSED(argv)
+
+    char cwd[PATH_MAX];
+    if (IS_NULL_PTR(getcwd(cwd, PATH_MAX))) {


PATH_MAX is not defined on STM32 and RPi Pico, also more broadly getcwd() is POSIX but not standard C, so I don't expect it to be available on all platforms.
We might use some kind of HAVE_CWD define to check if a platform has it or not, as we are already doing with open() and other POSIX functions.
Question: should we move this to posix_nifs.c? CC @pguyot

Indeed, if we test for HAVE_GETCWD because it's posix and not C, I'd say it makes sense to move it to posix_nifs.c.

bettio · 2025-09-07T17:41:46Z

src/libAtomVM/nifs.c

+    UNUSED(argc)
+    UNUSED(argv)
+
+    char cwd[PATH_MAX];


PATH_MAX might be quite big, I'm not sure this is safe for stack allocation.

bettio · 2025-09-07T17:43:20Z

src/libAtomVM/nifs.c

 #include <stdio.h>
 #include <string.h>
 #include <time.h>
+#include <unistd.h>


As I mention later, I'm worried we are leaking unistd and POSIX stuff into platform agnostic nifs.c.

bettio · 2025-09-07T17:44:31Z

tests/erlang_tests/test_file_get_cwd.erl

+    true = is_valid_path(Path),
+    ok.
+
+is_valid_path(<<"/", _/binary>>) ->


We'll regret this the day we are going to test this on a platform that is not unix ;)
Let's check platform.

bettio · 2025-09-19T20:23:44Z

tests/erlang_tests/test_file_get_cwd.erl

+% SPDX-License-Identifier: Apache-2.0 OR LGPL-2.1-or-later
+%
+
+-module(test_file_get_cwd).


I think we should have just a ŧest_file module as we are doing with other modules in our tests.
In order to get this done in a pain-free manner I suggest merging #1807 first, then we can rebase this, add relevant tests to test_file and squash everything.

bettio

Everything good, let's squash everything into a single commit.

mat-hek · 2025-09-24T09:13:18Z

@bettio done. BTW, can't you do squash and merge on github? Would save us one step ;)

bettio · 2025-09-24T13:48:03Z

src/libAtomVM/nifs.c


-static const struct Nif make_ref_nif = {
+static const struct Nif make_ref_nif =
+{


I didn't notice this. Let's rewrite this as:

static const struct Nif make_ref_nif = {

So clang-format check will pass.

bettio · 2025-09-24T13:55:34Z

src/libAtomVM/CMakeLists.txt

 define_if_function_exists(libAtomVM unlink "unistd.h" PRIVATE HAVE_UNLINK)
 define_if_function_exists(libAtomVM execve "unistd.h" PRIVATE HAVE_EXECVE)
 define_if_function_exists(libAtomVM closefrom "unistd.h" PRIVATE HAVE_CLOSEFROM)
+define_if_function_exists(libAtomVM getcwd "unistd.h" PUBLIC HAVE_GETCWD)


I suggest using define_if_symbol_exists for PATH_MAX: on Pico it is failing with the following error:

/home/runner/work/AtomVM/AtomVM/src/libAtomVM/posix_nifs.c: In function 'nif_file_get_cwd_0': /home/runner/work/AtomVM/AtomVM/src/libAtomVM/posix_nifs.c:953:24: error: 'PATH_MAX' undeclared (first use in this function) 953 | char *cwd = malloc(PATH_MAX); | ^~~~~~~~

So we need to define HAVE_PATH_MAX and we'll have the following:

#if defined(HAVE_GETCWD) && defined(HAVE_PATH_MAX) static term nif_file_get_cwd_0(Context *ctx, int argc, term argv[])

These changes are made under both the "Apache 2.0" and the "GNU Lesser General Public License 2.1 or later" license terms (dual license). SPDX-License-Identifier: Apache-2.0 OR LGPL-2.1-or-later Added nif implementation for getcwd. Signed-off-by: Mateusz Front <[email protected]>

mat-hek force-pushed the mf/prim-file-get-cwd branch 2 times, most recently from 19e7b57 to e2ae3ca Compare August 20, 2025 12:56

mat-hek changed the title ~~Add prim_file:get_cwd/0~~ Add file:get_cwd/0 Aug 22, 2025

bettio requested changes Sep 7, 2025

View reviewed changes

mat-hek force-pushed the mf/prim-file-get-cwd branch from 7692b6d to 980951e Compare September 16, 2025 10:14

mat-hek requested a review from bettio September 17, 2025 08:14

bettio requested changes Sep 19, 2025

View reviewed changes

mat-hek force-pushed the mf/prim-file-get-cwd branch 2 times, most recently from 53f1a4b to ea6d34e Compare September 23, 2025 10:00

mat-hek requested a review from bettio September 23, 2025 10:27

bettio requested changes Sep 23, 2025

View reviewed changes

mat-hek force-pushed the mf/prim-file-get-cwd branch 2 times, most recently from 898b61c to d580122 Compare September 24, 2025 09:12

mat-hek requested a review from bettio September 24, 2025 09:22

bettio requested changes Sep 24, 2025

View reviewed changes

mat-hek force-pushed the mf/prim-file-get-cwd branch 6 times, most recently from 9311e6e to 1075f7b Compare September 26, 2025 11:44

mat-hek force-pushed the mf/prim-file-get-cwd branch from 1075f7b to 8c2d206 Compare September 26, 2025 13:03

mat-hek requested a review from bettio September 26, 2025 13:03

bettio approved these changes Sep 26, 2025

View reviewed changes

bettio merged commit f4973f2 into atomvm:main Sep 26, 2025
128 of 130 checks passed

Add file:get_cwd/0 #1806

Add file:get_cwd/0 #1806

Uh oh!

Conversation

mat-hek commented Aug 20, 2025

Uh oh!

pguyot commented Aug 20, 2025

Uh oh!

mat-hek commented Aug 21, 2025

Uh oh!

pguyot commented Aug 21, 2025

Uh oh!

mat-hek commented Aug 22, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

bettio left a comment

Choose a reason for hiding this comment

Uh oh!

mat-hek commented Sep 24, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!