Implement mlock(), kind of.

Currently mlock() and friends do nothing whatsoever. However, mlocking
is directly application-visible in a number of ways; for example,
madvise(MADV_DONTNEED) and msync(MS_INVALIDATE) both fail on mlocked
regions. We handle this inconsistently: MADV_DONTNEED is too important
to not work, but MS_INVALIDATE is rejected.

Change MM to track mlocked regions in a manner consistent with Linux.
It still will not actually pin pages into host physical memory, but:

- mlock() will now cause sentry memory management to precommit mlocked
pages.

- MADV_DONTNEED and MS_INVALIDATE will interact with mlocked pages as
described above.

PiperOrigin-RevId: 225861605
Change-Id: Iee187204979ac9a4d15d0e037c152c0902c8d0ee

2 files changed, 13 insertions, 1 deletions

diff --git a/pkg/abi/linux/limits.go b/pkg/abi/linux/limits.go
index b2e51b9bd..e0aa5b31d 100644
--- a/pkg/abi/linux/limits.go
+++ b/pkg/abi/linux/limits.go
@@ -60,7 +60,7 @@ const (
 	DefaultNofileHardLimit = 4096
 
 	// DefaultMemlockLimit is called MLOCK_LIMIT in Linux.
-	DefaultMemlockLimit = 64 * 1094
+	DefaultMemlockLimit = 64 * 1024
 
 	// DefaultMsgqueueLimit is called MQ_BYTES_MAX in Linux.
 	DefaultMsgqueueLimit = 819200
diff --git a/pkg/abi/linux/mm.go b/pkg/abi/linux/mm.go
index 3fcdf8235..eda8d9788 100644
--- a/pkg/abi/linux/mm.go
+++ b/pkg/abi/linux/mm.go
@@ -49,6 +49,18 @@ const (
 	MREMAP_FIXED   = 1 << 1
 )
 
+// Flags for mlock2(2).
+const (
+	MLOCK_ONFAULT = 0x01
+)
+
+// Flags for mlockall(2).
+const (
+	MCL_CURRENT = 1
+	MCL_FUTURE  = 2
+	MCL_ONFAULT = 4
+)
+
 // Advice for madvise(2).
 const (
 	MADV_NORMAL       = 0