[LU-18776] DoM discard callback can lead to recursion and stack corruption

• Type: Bug
• Resolution: Fixed
• Priority: Major
• Fix Version/s: Lustre 2.17.0
• Severity: 3
• Assignee: Mikhail Pershin
• Reporter: Mikhail Pershin

Description

the mdt_dom_discard_data() has two problems to resolve:

1. no protection from multiple calls for the same object. That causes many discard locks to be taken waiting each other, while only one is needed
2. mdt_dom_discard_cb() may cause another call to cancel callback in the same thread, entering into recursion. With many waiting locks (see #1) that causes stack overflow