On 08/11/12 22:35, Nicolas Pitre wrote:
On Fri, 10 Aug 2012, Cyril Chemparathy wrote:

This patch adds basic sanity tests to ensure that the instruction patching
results in valid instruction encodings.  This is done by verifying the output
of the patch process against a vector of assembler generated instructions at
init time.

Signed-off-by: Cyril Chemparathy <cy...@ti.com>
---
[...]
+       __asm__ __volatile__ (
+               "  .irp    shift1, 0, 6, 12, 18\n"
+               "  .irp    shift2, 0, 1, 2, 3, 4, 5\n"
+               "  add     r1, r2, #(0x41 << (\\shift1 + \\shift2))\n"
+               "  .endr\n"
+               "  .endr\n"


Maybe adding a "add r1, r2 #0x81 << 24" here might be a good thing since
this is the most used case but missing from the above.


Indeed. I've now replaced this with something a bit more extensive. Does the following look better?

+struct patch_test_imm8 {
+       u16     imm;
+       u16     shift;
+       u32     insn;
+};
+
+static void __init __used __naked __patch_test_code_imm8(void)
+{
+       __asm__ __volatile__ (
+
+               /* a single test case */
+               "       .macro          test_one, imm, sft\n"
+               "       .hword          \\imm\n"
+               "       .hword          \\sft\n"
+               "       add             r1, r2, #(\\imm << \\sft)\n"
+               "       .endm\n"
+
+               /* a sequence of tests at 'inc' increments of shift */
+               "       .macro          test_seq, imm, sft, max, inc\n"
+               "       test_one        \\imm, \\sft\n"
+               "       .if             \\sft < \\max\n"
+ " test_seq \\imm, (\\sft + \\inc), \\max, \\inc\n"
+               "       .endif\n"
+               "       .endm\n"
+
+               /* an empty record to mark the end */
+               "       .macro          test_end\n"
+               "       .hword          0, 0\n"
+               "       .word           0\n"
+               "       .endm\n"
+
+               /* finally generate the test sequences */
+               "       test_seq        0x41, 0, 24, 1\n"
+               "       test_seq        0x81, 0, 24, 2\n"
+               "       test_end\n"
+               : : :
+       );
+}
+

[...]
+       u32 test_code_addr = (u32)(&__patch_test_code_imm8);
+       u32 *test_code = (u32 *)(test_code_addr & ~0x3);

Why this masking?  With Thumb2 you may find functions starting at
halfword aligned addresses.  Only the LSB (indicating thumb mode) should
be masked out.


Fixed.  Thanks.

Thanks
-- Cyril.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Reply via email to