[PATCH] tools: support decoding ARM EHABI opcodes in readobj

Logan Chien tzuhsiang.chien at gmail.com
Sun Jan 19 08:52:45 PST 2014


  Hi Saleem,

  IMO, we need more test cases for higher coverage.  For example, I have written some:


      	.globl	func1
      	.type	func1,%function
      	.align	2
      	.fnstart
      func1:
      	.unwind_raw 12, 0x02
      	.unwind_raw -12, 0x42
      	.unwind_raw 0, 0x80, 0x00
      	.unwind_raw 4, 0x81, 0x00
      	.unwind_raw 4, 0x80, 0x01
      	.unwind_raw 8, 0x80, 0xc0
      	.unwind_raw 12, 0x84, 0xc0
      	.unwind_raw 0, 0x91
      	.unwind_raw 8, 0xa1
      	.unwind_raw 12, 0xa9
      	.unwind_raw 0, 0xb0
      	.unwind_raw 4, 0xb1, 0x01
      	.unwind_raw 0xa04, 0xb2, 0x80, 0x4
      	.unwind_raw 24, 0xb3, 0x12
      	.unwind_raw 24, 0xba
      	.unwind_raw 24, 0xc2
      	.unwind_raw 24, 0xc6, 0x02
      	.unwind_raw 8, 0xc7, 0x03
      	.unwind_raw 24, 0xc8, 0x02
      	.unwind_raw 24, 0xc9, 0x02
      	.unwind_raw 64, 0xd7
      	.fnend

  It seems that some decoding logic is not correct.  Unfortunately, I haven't got time to debug this yet.  Please have a look on the test cases.  Thanks.

  Logan

http://llvm-reviews.chandlerc.com/D2565



More information about the llvm-commits mailing list