From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: To: bug-binutils@gnu.org Cc: parisc-linux@lists.parisc-linux.org Message-Id: <20030131070155.8DDFD482A@dsl2.external.hp.com> Date: Fri, 31 Jan 2003 00:01:55 -0700 (MST) From: grundler@dsl2.external.hp.com (Grant Grundler) Subject: [parisc-linux] PATCH hppa ordered load absolute ops Sender: parisc-linux-admin@lists.parisc-linux.org Errors-To: parisc-linux-admin@lists.parisc-linux.org List-Help: List-Post: List-Subscribe: , List-Id: parisc-linux developers list List-Unsubscribe: , List-Archive: in trying to close a potential memory ordering hole for pa2.0 machines, I've run into the following problem with binutils. hppa linux kernel uses ldwa to access MMIO space instead of using virtually mapped addresses. Normally, ordering is enforced automatically to all IO space accesses according to the PA 2.0 architecture. However, some platforms (not supported yet) have IO devices responding to physical addresses *outside* the architected IO spaces and that's the hole I was trying to close. hppa64-linux-gcc -D__KERNEL__ -I/home/src/2.4.20/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -fomit-frame-pointer -mpa-risc-2-0 -D__linux__ -pipe -fno-strength-reduce -mno-space-regs -mfast-indirect-calls -mdisable-fpregs -ffunction-sections -march=2.0 -mschedule=8000 -nostdinc -I /opt/palinux/lib/gcc-lib/hppa64-linux/3.0.4/include -DKBUILD_BASENAME=eeprom -c -o eeprom.o eeprom.c {standard input}: Assembler messages: {standard input}:1039: Error: Invalid operands {standard input}:1054: Error: Invalid operands ... the -S output from eeprom.s looks like: ... 1034 #NO_APP 1035 extrd,u %r19,63,32,%r19 1036 #APP 1037 stwa,o %r19,0(%r26) 1038 1039 ldwa,o 0(%r26),%r21 1040 1041 #NO_APP 1042 or %r20,%r22,%r20 As one can see, "stwa,o" was ok but "ldwa,o" was not. I've checked the pa2.0 arch and one of the restrictions is the "im5" value must be zero (which it obviously is). Appended is my very first attempt at adding a new opcode. i think I have the match/mask fields right. I'm still a bit sketchy on co/cc/@ specs. thanks, grant --- ./binutils-2.13.90.0.16/include/opcode/hppa.h-orig 2003-01-30 21:06:56.000000000 -0800 +++ ./binutils-2.13.90.0.16/include/opcode/hppa.h 2003-01-30 22:58:54.000000000 -0800 @@ -405,6 +405,7 @@ static const struct pa_opcode pa_opcodes { "ldbx", 0x0c000000, 0xfc001fc0, "cXx(b),t", pa10, 0}, { "ldwa", 0x0c000180, 0xfc00d3c0, "cxccx(b),t", pa10, FLAG_STRICT}, { "ldwa", 0x0c001180, 0xfc00d3c0, "cmcc5(b),t", pa10, FLAG_STRICT}, +{ "ldwa", 0x0c0031a0, 0xfc1ff3d0, "cocc@(b),t", pa20, FLAG_STRICT}, { "ldcw", 0x0c0001c0, 0xfc0013c0, "cxcdx(s,b),t", pa10, FLAG_STRICT}, { "ldcw", 0x0c0001c0, 0xfc0013c0, "cxcdx(b),t", pa10, FLAG_STRICT}, { "ldcw", 0x0c0011c0, 0xfc0013c0, "cmcd5(s,b),t", pa10, FLAG_STRICT}, @@ -415,6 +416,7 @@ static const struct pa_opcode pa_opcodes { "stby", 0x0c001300, 0xfc0013c0, "cscCx,V(b)", pa10, FLAG_STRICT}, { "ldda", 0x0c000100, 0xfc00d3c0, "cxccx(b),t", pa20, FLAG_STRICT}, { "ldda", 0x0c001100, 0xfc00d3c0, "cmcc5(b),t", pa20, FLAG_STRICT}, +{ "ldda", 0x0c003100, 0xfc1ff3d0, "cocc@(b),t", pa20, FLAG_STRICT}, { "ldcd", 0x0c000140, 0xfc0013c0, "cxcdx(s,b),t", pa20, FLAG_STRICT}, { "ldcd", 0x0c000140, 0xfc0013c0, "cxcdx(b),t", pa20, FLAG_STRICT}, { "ldcd", 0x0c001140, 0xfc0013c0, "cmcd5(s,b),t", pa20, FLAG_STRICT},