On 22/11/2023 22:35, Andrew Cooper wrote:
On 06/11/2023 3:05 pm, Alejandro Vallejo wrote:
Create a wrapper for the new fdopen() function of libfsimage.
Signed-off-by: Alejandro Vallejo <alejandro.vall...@cloud.com>
I'd appreciate it if Marek would cast his eye (as python maintainer)
over it.
That said, ...
diff --git a/tools/pygrub/src/fsimage/fsimage.c
b/tools/pygrub/src/fsimage/fsimage.c
index 12dfcff6e3..216f265331 100644
--- a/tools/pygrub/src/fsimage/fsimage.c
+++ b/tools/pygrub/src/fsimage/fsimage.c
@@ -270,6 +270,30 @@ fsimage_open(PyObject *o, PyObject *args, PyObject *kwargs)
return (PyObject *)fs;
}
+static PyObject *
+fsimage_fdopen(PyObject *o, PyObject *args, PyObject *kwargs)
+{
+ static char *kwlist[] = { "fd", "offset", "options", NULL };
+ int fd;
+ char *options = NULL;
+ uint64_t offset = 0;
+ fsimage_fs_t *fs;
+
+ if (!PyArg_ParseTupleAndKeywords(args, kwargs, "i|Ls", kwlist,
+ &fd, &offset, &options))
+ return (NULL);
+
+ if ((fs = PyObject_NEW(fsimage_fs_t, &fsimage_fs_type)) == NULL)
+ return (NULL);
+
+ if ((fs->fs = fsi_fdopen_fsimage(fd, offset, options)) == NULL) {
+ PyErr_SetFromErrno(PyExc_IOError);
Don't we need a Py_DECREF(fs) here to avoid leaking it?
~Andrew
If so, there's a bug in fsimage_open() as well. The logic here identical
to the logic there.
Cheers,
Alejandro
