pc64: track per process kmap page tables in separate MMU list.

we have to keep kmap page tables in ther own list
because user tables are subject to (virtual) tlb flushing.

we never free kmap page tables except in mmurelease()
where we just link the kmap mmu list in front of the user
mmus and call mmufree() which will free all the mmu's
of the process.

1 files changed, 45 insertions, 28 deletions

diff --git a/sys/src/9/pc64/mmu.c b/sys/src/9/pc64/mmu.c
index b30cf4815..1c406ecc6 100644
--- a/sys/src/9/pc64/mmu.c
+++ b/sys/src/9/pc64/mmu.c
@@ -227,25 +227,36 @@ mmuwalk(uintptr* table, uintptr va, int level, int create)
 				return 0;
 			pte = PTEWRITE|PTEVALID;
 			if(va < VMAP){
-				if(va < TSTKTOP)
+				if(va < TSTKTOP){
 					pte |= PTEUSER;
-				p = mmualloc();
-				p->index = x;
-				p->level = i;
-				if(i == PML4E){
-					/* PML4 entries linked to head */
-					p->next = up->mmuhead;
-					if(p->next == nil)
-						up->mmutail = p;
-					up->mmuhead = p;
-					if(p->index <= PTLX(TSTKTOP, 3))
+
+					p = mmualloc();
+					p->index = x;
+					p->level = i;
+					if(i == PML4E){
+						if((p->next = up->mmuhead) == nil)
+							up->mmutail = p;
+						up->mmuhead = p;
 						m->mmumap[p->index/MAPBITS] |= 1ull<<(p->index%MAPBITS);
-				} else {
-					/* PDP and PD entries linked to tail */
-					up->mmutail->next = p;
-					up->mmutail = p;
-				}
-				up->mmucount++;
+					} else {
+						up->mmutail->next = p;
+						up->mmutail = p;
+					}
+					up->mmucount++;
+				} else if(va >= KMAP && va < (KMAP+KMAPSIZE)) {
+					p = mmualloc();
+					p->index = x;
+					p->level = i;
+					if(i == PML4E){
+						up->kmaptail = p;
+						up->kmaphead = p;
+					} else {
+						up->kmaptail->next = p;
+						up->kmaptail = p;
+					}
+					up->kmapcount++;
+				} else
+					return 0;
 				page = p->page;
 			} else if(didmmuinit) {
 				page = mallocalign(PTSZ, BY2PG, 0, 0);
@@ -375,7 +386,6 @@ flushmmu(void)
 void
 mmuswitch(Proc *proc)
 {
-	uintptr pte;
 	MMU *p;
 
 	mmuzap();
@@ -383,13 +393,11 @@ mmuswitch(Proc *proc)
 		mmufree(proc);
 		proc->newtlb = 0;
 	}
-	for(p = proc->mmuhead; p && p->level==PML4E; p = p->next){
-		pte = PADDR(p->page) | PTEWRITE|PTEVALID;
-		if(p->index <= PTLX(TSTKTOP, 3)){
-			m->mmumap[p->index/MAPBITS] |= 1ull<<(p->index%MAPBITS);
-			pte |= PTEUSER;
-		}
-		m->pml4[p->index] = pte;
+	if((p = proc->kmaphead) != nil)
+		m->pml4[PTLX(KMAP, 3)] = PADDR(p->page) | PTEWRITE|PTEVALID;
+	for(p = proc->mmuhead; p != nil && p->level == PML4E; p = p->next){
+		m->mmumap[p->index/MAPBITS] |= 1ull<<(p->index%MAPBITS);
+		m->pml4[p->index] = PADDR(p->page) | PTEUSER|PTEWRITE|PTEVALID;
 	}
 	taskswitch((uintptr)proc->kstack+KSTACK);
 }
@@ -397,7 +405,18 @@ mmuswitch(Proc *proc)
 void
 mmurelease(Proc *proc)
 {
+	MMU *p;
+
 	mmuzap();
+	if((p = proc->kmaptail) != nil){
+		if((p->next = proc->mmuhead) == nil)
+			proc->mmutail = p;
+		proc->mmuhead = p;
+		proc->mmucount += proc->kmapcount;
+
+		proc->kmaphead = proc->kmaptail = nil;
+		proc->kmapcount = 0;
+	}
 	mmufree(proc);
 	taskswitch((uintptr)m+MACHSIZE);
 }
@@ -410,10 +429,8 @@ putmmu(uintptr va, uintptr pa, Page *)
 
 	x = splhi();
 	pte = mmuwalk(m->pml4, va, 0, 1);
-	if(pte == 0){
+	if(pte == 0)
 		panic("putmmu: bug: va=%#p pa=%#p", va, pa);
-		return;
-	}
 	old = *pte;
 	*pte = pa | PTEVALID|PTEUSER;
 	splx(x);