Re: [PATCH v3 2/2] libfdt: tests: add get_next_tag_invalid_prop_len

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]



On 10/6/22 01:06, David Gibson wrote:
On Wed, Oct 05, 2022 at 04:29:31PM -0700, Tadeusz Struk wrote:
Add a new test get_next_tag_invalid_prop_len, which covers
fdt_next_tag(), when it is passed an corrupted blob, with
invalid property len values.

Signed-off-by: Tadeusz Struk<tadeusz.struk@xxxxxxxxxx>
Looks good overall, but a bunch of minor things to polish.

Thanks for reviewing this. I will only follow with the new
version if 2/2. Please take the v3 1/2 as it is.


+#define FDT_SIZE 65536
+#define CHECK_ERR(err) \
+({ if (err) { \
+	free(fdt); \
You don't need the free() here, you're about to quit the test program
anyway.

Right. I will take that out.


+	FAIL("%s: %d: %s", __FILE__, __LINE__, fdt_strerror(err)); \
+	} \
+})
+
+int main(int argc, char *argv[])
+{
+	struct fdt_property *prp;
+	void *fdt;
+	int nextoff = 0, offset, err;
+	uint32_t tag, val;
+
+	test_init(argc, argv);
+	fdt = calloc(1, FDT_SIZE);
No need to use cleared memory, the fdt_sw functions will work just
fine with an uninitialized buffer.

Ok


+	if (!fdt)
+		FAIL("Can't allocate memory");
+	err = fdt_create(fdt, FDT_SIZE);
+	CHECK_ERR(err);
+	err = fdt_add_reservemap_entry(fdt, 0xdeadbeefUL, 0x10000UL);> No need to insert a dummy reservemap entry here.

Ok, removed


+	CHECK_ERR(err);
+	err = fdt_finish_reservemap(fdt);
+	CHECK_ERR(err);
+	err = fdt_begin_node(fdt, "");
+	CHECK_ERR(err);
+	err = fdt_begin_node(fdt, "subnode1");
+	CHECK_ERR(err);
No particular need for this subnode either, you can test what you want
with properties on the root node.

Removed the extra subnode.


+	err = fdt_property_u32(fdt, "prop-int-32", 0x1234);
+	CHECK_ERR(err);
+	err = fdt_property_u32(fdt, "prop2-int-32", 0x4321);
+	CHECK_ERR(err);
+	err = fdt_end_node(fdt);
+	CHECK_ERR(err);
+	err = fdt_end_node(fdt);
+	CHECK_ERR(err);
+	offset = -1;
+	val = cpu_to_fdt32(0x1234);
+	offset = fdt_node_offset_by_prop_value(fdt, offset,  "prop-int-32",
+					       &val, sizeof(val));
fdt_node_offset_by_prop_value() is a very roundabout way to find the
node you need - you know the path (and if you get rid of the
unnecessary subnode, it'll just be the root node at offset 0).

I removed that as well.


+	do {
+		tag = fdt_next_tag(fdt, offset, &nextoff);
+		offset = nextoff;
+	} while (tag != FDT_PROP);
+
+	/* Calculate len to property */
+	prp = (struct fdt_property *)(((char*)fdt) + fdt_off_dt_struct(fdt) + offset);
You could replace the loop as well as this nasty pointer arithmetic
with an fdt_get_property_w() call.

The fdt_get_property_w() was what I was looking for, thanks, but even using it
I'm getting a different pointer within the fdt to what I'm getting when I
calculate the offset myself. I the tag value that it is pointing to is the
FDT_BEGIN_NODE. Do I need to do anything else after with the pointer returned
from fdt_get_property_w()?


+	/* int overflow case */
Probably worth testing the fdt_next_tag() behaviour on the unmangled
tree before testing the corrupted cases.  If the test ever breaks,
that sort of thing makes it easier to figure out if the breakage is in
the library, or the testcase.
Will do.

--
Thanks,
Tadeusz




[Index of Archives]     [Device Tree]     [Device Tree Spec]     [Linux Driver Backports]     [Video for Linux]     [Linux USB Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]

  Powered by Linux