Date: Thu, 22 Oct 2015 11:22:58 +0000 From: bugzilla-noreply@freebsd.org To: freebsd-arm@FreeBSD.org Subject: [Bug 203950] clang mis-assembles immediate operands Message-ID: <bug-203950-7@https.bugs.freebsd.org/bugzilla/>
next in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=203950 Bug ID: 203950 Summary: clang mis-assembles immediate operands Product: Base System Version: 11.0-CURRENT Hardware: Any OS: Any Status: New Severity: Affects Only Me Priority: --- Component: arm Assignee: freebsd-arm@FreeBSD.org Reporter: peterj@FreeBSD.org Created attachment 162344 --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=162344&action=edit ARM AES code generated by OpenSSL perl script. Whilst experimenting with the OpenSSL ARM assembler code, I discovered that clang is incorrectly handling immediate operands. This is using: % cc -c aes_arm.s % objdump -d aes_arm.o | less % cc -v FreeBSD clang version 3.6.1 (tags/RELEASE_361/final 237755) 20150525 Target: armv6--freebsd11.0-gnueabihf Thread model: posix The source code (attached) includes: AES_encrypt: sub r3,pc,#8 @ AES_encrypt stmdb sp!,{r1,r4-r12,lr} mov r12,r0 @ inp mov r11,r2 sub r10,r3,#AES_encrypt-AES_Te @ Te #AES_encrypt-AES_Te is 1344 or 0x540. clang compiles this subtraction to 0xe243a540 - which decompiles to: e243a540 sub sl, r3, #268435456 ; 0x10000000 because the 12-bit immediate field is not a direct binary number but is a 4-bit rotation followed by a 8-bit binary value. A correct value for this field would be (eg) 0xe54. Note that as(1) gets compiles the instruction to: e243ad15 sub sl, r3, #1344 ; 0x540 -- You are receiving this mail because: You are the assignee for the bug.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-203950-7>