Index: head/sys/boot/i386/boot2/Makefile
===================================================================
--- head/sys/boot/i386/boot2/Makefile	(revision 316063)
+++ head/sys/boot/i386/boot2/Makefile	(revision 316064)
@@ -1,116 +1,124 @@
 # $FreeBSD$
 
 .include <bsd.own.mk>
 
 FILES=		boot boot1 boot2
 
 NM?=		nm
 
 # A value of 0x80 enables LBA support.
 BOOT_BOOT1_FLAGS?=	0x80
 
 BOOT_COMCONSOLE_PORT?= 0x3f8
 BOOT_COMCONSOLE_SPEED?= 9600
 B2SIOFMT?=	0x3
 
 REL1=	0x700
 ORG1=	0x7c00
 ORG2=	0x2000
 
 # Decide level of UFS support.
 BOOT2_UFS?=	UFS1_AND_UFS2
 #BOOT2_UFS?=	UFS2_ONLY
 #BOOT2_UFS?=	UFS1_ONLY
 
 CFLAGS=	-fomit-frame-pointer \
 	-mrtd \
 	-mregparm=3 \
 	-D${BOOT2_UFS} \
 	-DFLAGS=${BOOT_BOOT1_FLAGS} \
 	-DSIOPRT=${BOOT_COMCONSOLE_PORT} \
 	-DSIOFMT=${B2SIOFMT} \
 	-DSIOSPD=${BOOT_COMCONSOLE_SPEED} \
 	-I${.CURDIR}/../../common \
 	-I${.CURDIR}/../btx/lib -I. \
 	-Wall -Waggregate-return -Wbad-function-cast -Wno-cast-align \
 	-Wmissing-declarations -Wmissing-prototypes -Wnested-externs \
 	-Wpointer-arith -Wshadow -Wstrict-prototypes -Wwrite-strings \
 	-Winline
 
 CFLAGS.gcc+=	-Os \
 		-fno-guess-branch-probability \
 		-fno-unit-at-a-time \
 		--param max-inline-insns-single=100
 .if ${COMPILER_TYPE} == "gcc" && ${COMPILER_VERSION} <= 40201
 CFLAGS.gcc+=   -mno-align-long-strings
 .endif
 
 CFLAGS.clang+=	-Oz ${CLANG_OPT_SMALL}
 
 LD_FLAGS=${LD_FLAGS_BIN}
 
 # Pick up ../Makefile.inc early.
 .include <bsd.init.mk>
 
 CLEANFILES=	boot
 
 boot: boot1 boot2
 	cat boot1 boot2 > boot
 
 CLEANFILES+=	boot1 boot1.out boot1.o
 
 boot1: boot1.out
 	${OBJCOPY} -S -O binary boot1.out ${.TARGET}
 
 boot1.out: boot1.o
 	${LD} ${LD_FLAGS} -e start -Ttext ${ORG1} -o ${.TARGET} boot1.o
 
 CLEANFILES+=	boot2 boot2.ld boot2.ldr boot2.bin boot2.out boot2.o \
 		boot2.s boot2.s.tmp boot2.h sio.o
 
 boot2: boot2.ld
 	@set -- `ls -l boot2.ld`; x=$$((7680-$$5)); \
 	    echo "$$x bytes available"; test $$x -ge 0
 	${DD} if=boot2.ld of=${.TARGET} obs=7680 conv=osync
 
 boot2.ld: boot2.ldr boot2.bin ${BTXKERN}
 	btxld -v -E ${ORG2} -f bin -b ${BTXKERN} -l boot2.ldr \
 	    -o ${.TARGET} -P 1 boot2.bin
 
 boot2.ldr:
 	${DD} if=/dev/zero of=${.TARGET} bs=512 count=1
 
 boot2.bin: boot2.out
 	${OBJCOPY} -S -O binary boot2.out ${.TARGET}
 
 boot2.out: ${BTXCRT} boot2.o sio.o
 	${LD} ${LD_FLAGS} -Ttext ${ORG2} -o ${.TARGET} ${.ALLSRC}
 
 boot2.o: boot2.s
 	${CC} ${ACFLAGS} -c boot2.s
 
 SRCS=	boot2.c boot2.h
 
+# Gcc (4.2.1 at least) benefits from removing the forced alignment
+# clang doesn't. Make the removal as specific as possible to avoid
+# false positives (like path names with odd names for debugging info).
+# In the past, gcc benefited from nop removal, but not in 4.2.1.
+# Think of this as a poor-man's peephole optimizer for gcc 4.2.1
 boot2.s: boot2.c boot2.h ${.CURDIR}/../../common/ufsread.c
 	${CC} ${CFLAGS} -S -o boot2.s.tmp ${.CURDIR}/boot2.c
-	sed -e '/align/d' -e '/nop/d' < boot2.s.tmp > boot2.s
-	rm -f boot2.s.tmp
+.if ${COMPILER_TYPE} == "gcc"
+	sed -e '/\.align 4/d' < boot2.s.tmp > boot2.s
+.else
+	cp boot2.s.tmp boot2.s
+.endif
 
 boot2.h: boot1.out
 	${NM} -t d ${.ALLSRC} | awk '/([0-9])+ T xread/ \
 	    { x = $$1 - ORG1; \
 	    printf("#define XREADORG %#x\n", REL1 + x) }' \
 	    ORG1=`printf "%d" ${ORG1}` \
 	    REL1=`printf "%d" ${REL1}` > ${.TARGET}
 
 .if ${MACHINE_CPUARCH} == "amd64"
 beforedepend boot2.s: machine
 CLEANFILES+=	machine
 machine: ${.CURDIR}/../../../i386/include .NOMETA
 	ln -sf ${.ALLSRC} ${.TARGET}
 .endif
 
 .include <bsd.prog.mk>
 
 # XXX: clang integrated-as doesn't grok .codeNN directives yet
 CFLAGS.boot1.S=		${CLANG_NO_IAS}