From f876e66e717c4dd853da12892048262afe47ffdf Mon Sep 17 00:00:00 2001 From: Alex Auvolat Date: Thu, 19 Feb 2015 20:18:39 +0100 Subject: Add syscall for ioctl on open handle --- src/kernel/include/vfs.h | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) (limited to 'src/kernel/include/vfs.h') diff --git a/src/kernel/include/vfs.h b/src/kernel/include/vfs.h index 37b28b9..bae4e6f 100644 --- a/src/kernel/include/vfs.h +++ b/src/kernel/include/vfs.h @@ -18,10 +18,15 @@ // some data managed by the underlying filesystem. The following types are aliases to void*, // but are used to disambiguate the different types of void* : fs_handle_ptr, fs_node_ptr, fs_ptr +// Conventions : +// - ioctl returns negative values on error. null or positives are valid return values from underlying +// driver + // About thread safety : -// - The VFS implements a locking mechanism on handles so that only one operation is executed -// at the same time on a given handle -// - Same for FS nodes +// - The VFS implements a locking mechanism on FS nodes so that only one operation is executed +// at the same time on a given node (including: open, stat, walk, delete, move, create, ioctl) +// - The VFS does not implement locks on handles : several simultaneous read/writes are possible, and +// it is the driver's responsiblity to prevent that if necessary // - The VFS does not lock nodes that have a handle open to them : it is the FS code's responsibility // to refuse some commands if neccessary on a node that is open. // - The VFS does not implement any locking mechanism on filesystems themselves (e.g. the add_source @@ -166,6 +171,7 @@ int file_get_mode(fs_handle_t *f); size_t file_read(fs_handle_t *f, size_t offset, size_t len, char* buf); size_t file_write(fs_handle_t *f, size_t offset, size_t len, const char* buf); bool file_stat(fs_handle_t *f, stat_t *st); +int file_ioctl(fs_handle_t *f, int command, void* data); bool file_readdir(fs_handle_t *f, dirent_t *d); /* vim: set ts=4 sw=4 tw=0 noet :*/ -- cgit v1.2.3