proc: fix gs/fs-related bug in exception handler

1. %rip -> %rsp (misspelt a register name, damn)
2. Rename swapgs_if_needed to iret_swapgs_if_needed for clarification
3. Add iret_swapgs_if_needed before iret from exception handler

arch/amd64/hw/exc_s.S

@@ -22,8 +22,8 @@ amd64_exc_generic:
     // 0x10: rip
     // 0x08: error code
     // 0x00: error number
-    cmpq $0x08, 0x18(%rip)
-    jnz 1f
+    cmpq $0x08, 0x18(%rsp)
+    jz 1f
     swapgs
 1:
 
@@ -79,6 +79,7 @@ amd64_exc_generic:
 
     addq $16, %rsp
 
+    iret_swapgs_if_needed
     iretq
 
 amd64_isr_nerr 0

arch/amd64/hw/irq0.S

@@ -7,16 +7,16 @@
 
 amd64_irq0_early:
     cli
-    swapgs_if_needed
+    iret_swapgs_if_needed
     pushq %rax
     irq_eoi_lapic 0
     popq %rax
-    swapgs_if_needed
+    iret_swapgs_if_needed
     iretq
 
 amd64_irq0:
     cli
-    swapgs_if_needed
+    iret_swapgs_if_needed
 
     // Push caller-saved registers so it appears as if a thread just called yield()
     pushq %r11
@@ -42,7 +42,7 @@ amd64_irq0:
     popq %r10
     popq %r11
 
-    swapgs_if_needed
+    iret_swapgs_if_needed
     iretq
 
 amd64_irq2_counting:

arch/amd64/hw/irqs_s.S

@@ -50,7 +50,7 @@ amd64_idt_load:
 .global amd64_irq_msi0
 amd64_irq_msi0:
     cli
-    swapgs_if_needed
+    iret_swapgs_if_needed
 
     pushq %r11
     pushq %r10
@@ -78,5 +78,5 @@ amd64_irq_msi0:
     popq %r10
     popq %r11
 
-    swapgs_if_needed
+    iret_swapgs_if_needed
     iretq

arch/amd64/sched_s.S

@@ -14,7 +14,7 @@ context_enter:
     popq %r10
     popq %r11
 
-    swapgs_if_needed
+    iret_swapgs_if_needed
     iretq
 
 .global context_exec_enter
@@ -53,7 +53,7 @@ context_exec_enter:
     xorq %r14, %r14
     xorq %r15, %r15
 
-    swapgs_if_needed
+    iret_swapgs_if_needed
     iretq
 
 .global context_switch_first

etc/Kernel.makefile

@@ -170,8 +170,7 @@ KERNEL_CFLAGS=-Iinclude \
 			  -Wall \
 			  -Wextra \
 			  -Wno-unused \
-			  -O2 \
-			  -funroll-loops \
+			  -Og \
 			  -ggdb \
 			  -Werror $(KERNEL_DEF)
 KERNEL_LDFLAGS=-nostdlib \

include/arch/amd64/asm/asm_cpu.h

@@ -11,9 +11,9 @@
     (__amd64_cpu + off)(%rip)
 #endif
 
-// To be used in interrupt handlers
+// To be used in interrupt handlers or anything with iret-like stack layout
 // DIRECTLY AFTER ENTRY without prior stack modification
-.macro swapgs_if_needed
+.macro iret_swapgs_if_needed
     cmpq $0x08, 0x08(%rsp)
     je 1f
     swapgs

include/arch/amd64/asm/asm_irq.h

@@ -13,7 +13,7 @@
 .global amd64_irq\n
 amd64_irq\n:
     cli
-    swapgs_if_needed
+    iret_swapgs_if_needed
 
     pushq %r11
     pushq %r10
@@ -39,7 +39,7 @@ amd64_irq\n:
     popq %r10
     popq %r11
 
-    swapgs_if_needed
+    iret_swapgs_if_needed
     iretq
 .endm