[LibOS] Rewrite rename* syscalls

[boryspoplawski]

Affected components

• README and global configuration
• Linux PAL
• SGX PAL
• FreeBSD PAL
• Common PAL code
• Library OS (i.e., SHIM), including GLIBC

Description of the changes

Unfortunately current implementation of fs does not allow for renaming anything but regular files.
Fixes some issues from #903 constrained by above statement.

How to test this PR?

---

This change is 

[yamahata]

The logic seems racy. But other fs related codes (open, unlink, rmdir, mkdir) seem racy.
So I don't block this due to it.

Reviewable status: 0 of 9 files reviewed, all discussions resolved, not enough approvals from maintainers (3 more required), not enough approvals from different teams (2 more required, approved so far: )

---

LibOS/shim/src/fs/shim_namei.c, line 834 at r1 (raw file):

}

int get_dirfd_dentry(int dirfd, struct shim_dentry** dir) {

Except the last put_handle(), Is this just code cleanup?

---

LibOS/shim/src/fs/shim_namei.c, line 858 at r1 (raw file):

    get_dentry(hdl->dentry);
    *dir = hdl->dentry;
    put_handle(hdl);

good catch.

---

LibOS/shim/src/sys/shim_fs.c, line 605 at r1 (raw file):

        }

        if ((ret = new_dent->fs->d_ops->unlink(new_dent->parent,

+1 to drop unlink logic. It should be up to fs.

---

LibOS/shim/src/sys/shim_fs.c, line 622 at r1 (raw file):

        new_dent->state &= ~DENTRY_NEGATIVE;
        if (old_dent->state & DENTRY_ISDIRECTORY) {
            new_dent->state |= DENTRY_ISDIRECTORY;

Is there any reason why this isn't updated before rename method?

[yamahata]

#944

Reviewed 9 of 9 files at r1.
Reviewable status: all files reviewed, 1 unresolved discussion, not enough approvals from maintainers (2 more required), not enough approvals from different teams (1 more required, approved so far: Intel) (waiting on @boryspoplawski)

[boryspoplawski]

I agree, but there are many more issues with current fs implementation and I don't think all of them can be fixed without major changes/reworks.

Reviewable status: all files reviewed, 1 unresolved discussion, not enough approvals from maintainers (2 more required), not enough approvals from different teams (1 more required, approved so far: Intel) (waiting on @yamahata)

---

LibOS/shim/src/fs/shim_namei.c, line 834 at r1 (raw file):

Previously, yamahata wrote…

Except the last put_handle(), Is this just code cleanup?

Yes, just wanted to change this function name, as previous was a bit missleading

---

LibOS/shim/src/sys/shim_fs.c, line 605 at r1 (raw file):

Previously, yamahata wrote…

+1 to drop unlink logic. It should be up to fs.

Exactly

---

LibOS/shim/src/sys/shim_fs.c, line 622 at r1 (raw file):

Previously, yamahata wrote…

Is there any reason why this isn't updated before rename method?

If rename fails, old_dent should stay in the same state as it was before.

[boryspoplawski]

Reviewable status: 8 of 9 files reviewed, 1 unresolved discussion, not enough approvals from maintainers (3 more required), not enough approvals from different teams (2 more required, approved so far: ) (waiting on @yamahata)

---

LibOS/shim/src/sys/shim_fs.c, line 622 at r1 (raw file):

Previously, boryspoplawski (Borys Popławski) wrote…

If rename fails, old_dent should stay in the same state as it was before.

Actually you are right. It was even updated above - so removed unneeded update here.

[yamahata]

implementation and I don't think all of them can be fixed without major changes/reworks.

fully agreed.

Reviewable status: 8 of 9 files reviewed, all discussions resolved, not enough approvals from maintainers (2 more required), not enough approvals from different teams (1 more required, approved so far: Intel) (waiting on @yamahata)

[mkow]

Reviewed 8 of 9 files at r1, 1 of 1 files at r2.
Reviewable status: all files reviewed, all discussions resolved, not enough approvals from maintainers (1 more required)

[mkow]

Jenkins, test this please

[mkow]

Reviewable status: all files reviewed, all discussions resolved, not enough approvals from maintainers (1 more required), not enough approvals from different teams (1 more required, approved so far: ITL)

[yamahata]

Reviewed 1 of 1 files at r2.
Reviewable status: complete! all files reviewed, all discussions resolved