diff mbox series

[v4,14/23] sandbox: Unmap old tags

Message ID 20240901222634.460873-15-sjg@chromium.org
State New
Delegated to: Tom Rini
Headers show
Series Fix various bugs | expand

Commit Message

Simon Glass Sept. 1, 2024, 10:26 p.m. UTC 
So far unmapping has not been implemented. This means that if one test
maps a pointer to an address with map_sysmem(), then a second test can
use that same pointer, by mapping the address back to a pointer with
map_to_sysmem(). This is not really desirable, even if it doesn't
cause any problems at the moment.

Implement unmapping, to clean this up.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 arch/sandbox/cpu/cpu.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
diff mbox series

Patch

diff --git a/arch/sandbox/cpu/cpu.c b/arch/sandbox/cpu/cpu.c
index 8eb055d6837..3e1c0dd583e 100644
--- a/arch/sandbox/cpu/cpu.c
+++ b/arch/sandbox/cpu/cpu.c
@@ -185,12 +185,28 @@  void *map_physmem(phys_addr_t paddr, unsigned long len, unsigned long flags)
 
 void unmap_physmem(const void *ptr, unsigned long flags)
 {
+	struct sandbox_mapmem_entry *mentry;
+
 #ifdef CONFIG_PCI
 	if (map_dev) {
 		pci_unmap_physmem(ptr, map_len, map_dev);
 		map_dev = NULL;
 	}
 #endif
+
+	/* If it is in emulated RAM, we didn't create a tag, so nothing to do */
+	if (is_in_sandbox_mem(ptr))
+		return;
+
+	mentry = find_tag(ptr);
+	if (mentry) {
+		list_del(&mentry->sibling_node);
+		log_debug("Removed map from %p to %lx\n", ptr,
+			  (ulong)mentry->tag);
+		free(mentry);
+	} else {
+		log_warning("Address not mapped: %p\n", ptr);
+	}
 }
 
 phys_addr_t map_to_sysmem(const void *ptr)