[NaCl SDK] nacl_io html5fs: Malformed file paths set errno to ENOENT.

[NaCl SDK] nacl_io html5fs: Malformed file paths set errno to ENOENT.

Applies to Open, Mkdir, Rmdir and Remove. Previously, these set errno to ENOSYS
or EIO.

BUG=252064
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=207856

[binji, 2013-06-20 15:56:12]

lgtm

(just to be clear here, though -- this is an error if the path is malformed, not
a missing file. PPB_FileRef will allow you to create a path to a non-existent
file/directory.)

https://codereview.chromium.org/17450014/diff/1/native_client_sdk/src/librari...
File native_client_sdk/src/libraries/nacl_io/mount_html5fs.cc (right):

https://codereview.chromium.org/17450014/diff/1/native_client_sdk/src/librari...
native_client_sdk/src/libraries/nacl_io/mount_html5fs.cc:62: return EIO;
change this one too? It's not immediately clear what the correct error to return
is (because we don't know why the FileRef create failed) but ENOENT (or ENOTDIR)
is probably better than EIO here.

[Matt Giuca, 2013-06-20 22:58:08]

On 2013/06/20 15:56:12, binji wrote:
> lgtm
> 
> (just to be clear here, though -- this is an error if the path is malformed,
not
> a missing file. PPB_FileRef will allow you to create a path to a non-existent
> file/directory.)

Oh, then I think this is all a misunderstanding caused by my bad mock in
https://codereview.chromium.org/15800004/. The folly of mocks!

I'm pretty sure in the real world case where the file is not found, it would
actually return ENOENT here: you're right, Create would succeed but Init would
fail with PP_ERROR_FILENOTFOUND which will translate into ENOENT.

So the bug I reported does not exist. Sorry. However, I think we both agree that
this should still be changed so that a malformed path gives ENOENT instead of
ENOSYS.

[Matt Giuca, 2013-06-21 00:33:03]

https://codereview.chromium.org/17450014/diff/1/native_client_sdk/src/librari...
File native_client_sdk/src/libraries/nacl_io/mount_html5fs.cc (right):

https://codereview.chromium.org/17450014/diff/1/native_client_sdk/src/librari...
native_client_sdk/src/libraries/nacl_io/mount_html5fs.cc:62: return EIO;
Yeah it's not quite clear to me what it should be either, but ENOENT seems
appropriate.

[commit-bot: I haz the power, 2013-06-21 00:34:36]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/mgiuca@chromium.org/17450014/6001

[commit-bot: I haz the power, 2013-06-21 16:31:52]

Change committed as 207856