Repository navigation
RLIMIT_AS and RLIMIT_DATA totals lose records once the region table is stale #488
Description
After #485, RLIMIT_AS and RLIMIT_DATA are checked against totals summed from the region table on each call. Once regions_tracker_stale is set, the records the table dropped stay out of both totals for good, so a guest that fills the table to GUEST_MAX_REGIONS and then splits mappings can grow past either limit. Running vm_total and vm_data counters, updated wherever a mapping is added, removed or reprotected and kept even when the table has no free slot, would close that gap and also replace the full-table scan on every mmap under a finite limit.
Activity
xalestar commentedon Nov 7, 2026
From the review of #485: the scan also costs time. With any finite limit, every
mmap,mremapgrowth and page-addingbrkwalks the whole region table, andMAP_FIXEDwalks it twice (mem_mapped_in, thenmem_vm_totals). A guest underulimit -vwith thousands of mappings (a JVM, a browser) then pays O(n) per call. Runningtotalanddatacounters kept in the region add, remove and prot-change paths would make this O(1), asmm->total_vmdoes, and the full sum could stay as a debug-build cross-check.