powerpc64: Use medium code model in asm files for TOC references

Summary:
With a sufficiently large TOC, it's possible to index out of range, as
the immediate load instructions only permit 16-bit indices, allowing up
to 64kB range (signed) from the base pointer.  Allow +/- 2GB range, with
the medium code model TOC accesses in asm.

Patch originally by Brandon Bergren.  The issue appears to impact ELFv2
more than ELFv1.

Reviewed by:	luporl
Differential Revision: https://reviews.freebsd.org/D19708

sys/powerpc/aim/trap_subr64.S

@@ -345,7 +345,8 @@ CNAME(rstcodeend):
 cpu_reset_handler:
 	GET_TOCBASE(%r2)
 
-	ld	%r1,TOC_REF(tmpstk)(%r2)	/* get new SP */
+	addis	%r1,%r2,TOC_REF(tmpstk)@ha
+	ld	%r1,TOC_REF(tmpstk)@l(%r1)	/* get new SP */
 	addi	%r1,%r1,(TMPSTKSZ-48)
 
 	bl	CNAME(cpudep_ap_early_bootstrap) /* Set PCPU */
@@ -380,7 +381,8 @@ cpu_wakeup_handler:
 	GET_TOCBASE(%r2)
 
 	/* Check for false wake up due to badly SRR1 set (eg. by OPAL) */
-	ld	%r3,TOC_REF(can_wakeup)(%r2)
+	addis	%r3,%r2,TOC_REF(can_wakeup)@ha
+	ld	%r3,TOC_REF(can_wakeup)@l(%r3)
 	ld	%r3,0(%r3)
 	cmpdi	%r3,0
 	beq	cpu_reset_handler
@@ -897,7 +899,8 @@ dbtrap:
 	mtsprg3	%r1
 
 	GET_TOCBASE(%r1)			/* get new SP */
-	ld	%r1,TOC_REF(trapstk)(%r1)
+	addis	%r1,%r1,TOC_REF(trapstk)@ha
+	ld	%r1,TOC_REF(trapstk)@l(%r1)
 	addi	%r1,%r1,(TRAPSTKSZ-48)
 
 	FRAME_SETUP(PC_DBSAVE)

sys/powerpc/booke/locore.S

@@ -303,7 +303,8 @@ done_mapping:
 	subf    %r31,%r31,%r2	/* Subtract from real TOC base to get base */
 
 	/* Set up the stack pointer */
-	ld	%r1,TOC_REF(tmpstack)(%r2)
+	addis	%r1,%r2,TOC_REF(tmpstack)@ha
+	ld	%r1,TOC_REF(tmpstack)@l(%r1)
 	addi	%r1,%r1,TMPSTACKSZ-96
 	add	%r1,%r1,%r31
 	bl	1f
@@ -552,7 +553,8 @@ bp_kernload:
 	mtspr	SPR_SPRG8, %r2
 
 	/* Set up the stack pointer */
-	ld	%r1,TOC_REF(tmpstack)(%r2)
+	addis	%r1,%r2,TOC_REF(tmpstack)@ha
+	ld	%r1,TOC_REF(tmpstack)@l(%r1)
 	addi	%r1,%r1,TMPSTACKSZ-96
 #else
 /*

sys/powerpc/ofw/ofwcall64.S

@@ -101,11 +101,13 @@ ASENTRY_NOPROF(ofwcall)
 	mfmsr	%r6
 
 	/* read client interface handler */
-	ld	%r4,TOC_REF(openfirmware_entry)(%r2)
+	addis	%r4,%r2,TOC_REF(openfirmware_entry)@ha
+	ld	%r4,TOC_REF(openfirmware_entry)@l(%r4)
 	ld	%r4,0(%r4)
 
 	/* Get OF stack pointer */
-	ld	%r7,TOC_REF(ofwstk)(%r2)
+	addis	%r7,%r2,TOC_REF(ofwstk)@ha
+	ld	%r7,TOC_REF(ofwstk)@l(%r7)
 	addi	%r7,%r7,OFWSTKSZ-40
 
 	/*
@@ -113,7 +115,8 @@ ASENTRY_NOPROF(ofwcall)
 	 * exceptions, which is important for the next few steps.
 	 */
 
-	ld	%r5,TOC_REF(ofmsr)(%r2)
+	addis	%r5,%r2,TOC_REF(ofmsr)@ha
+	ld	%r5,TOC_REF(ofmsr)@l(%r5)
 	ld	%r5,0(%r5)
 	mtmsrd	%r5
 	isync
@@ -233,16 +236,19 @@ ASENTRY_NOPROF(rtascall)
 	mfmsr	%r6
 
 	/* Read RTAS entry and reg save area pointers */
-	ld	%r5,TOC_REF(rtas_entry)(%r2)
+	addis	%r5,%r2,TOC_REF(rtas_entry)@ha
+	ld	%r5,TOC_REF(rtas_entry)@l(%r5)
 	ld	%r5,0(%r5)
-	ld	%r8,TOC_REF(rtas_regsave)(%r2)
+	addis	%r8,%r2,TOC_REF(rtas_regsave)@ha
+	ld	%r8,TOC_REF(rtas_regsave)@l(%r8)
 
 	/*
 	 * Set the MSR to the RTAS value. This has the side effect of disabling
 	 * exceptions, which is important for the next few steps.
 	 */
 
-	ld	%r7,TOC_REF(rtasmsr)(%r2)
+	addis	%r7,%r2,TOC_REF(rtasmsr)@ha
+	ld	%r7,TOC_REF(rtasmsr)@l(%r7)
 	ld	%r7,0(%r7)
 	mtmsrd	%r7
 	isync

sys/powerpc/powernv/opalcall.S

@@ -53,7 +53,8 @@ ASENTRY(opal_call)
 
 	/* Load OPAL entry information */
 	mr	%r0,%r3
-	ld	%r3,TOC_REF(opal_entrypoint)(%r2)
+	addis	%r3,%r2,TOC_REF(opal_entrypoint)@ha
+	ld	%r3,TOC_REF(opal_entrypoint)@l(%r3)
 	ld	%r3,0(%r3)
 	mtctr	%r3
 
@@ -62,9 +63,11 @@ ASENTRY(opal_call)
 	mfmsr	%r31
 
 	/* Load last bits from the TOC */
-	ld	%r3,TOC_REF(opal_msr)(%r2)
+	addis	%r3,%r2,TOC_REF(opal_msr)@ha
+	ld	%r3,TOC_REF(opal_msr)@l(%r3)
 	ld	%r3,0(%r3)
-	ld	%r2,TOC_REF(opal_data)(%r2)
+	addis	%r2,%r2,TOC_REF(opal_data)@ha
+	ld	%r2,TOC_REF(opal_data)@l(%r2)
 	ld	%r2,0(%r2)
 
 	mtmsrd	%r3

sys/powerpc/powerpc/swtch64.S

@@ -159,7 +159,8 @@ ENTRY(cpu_switch)
 cpu_switchin:
 #if defined(SMP) && defined(SCHED_ULE)
 	/* Wait for the new thread to become unblocked */
-	ld	%r6,TOC_REF(blocked_lock)(%r2)
+	addis	%r6,%r2,TOC_REF(blocked_lock)@ha
+	ld	%r6,TOC_REF(blocked_lock)@l(%r6)
 blocked_loop:
 	ld	%r7,TD_LOCK(%r13)
 	cmpd	%r6,%r7