summaryrefslogtreecommitdiff
path: root/arch/x86/kernel
diff options
context:
space:
mode:
authorPeter Zijlstra <peterz@infradead.org>2019-05-08 00:25:54 +0300
committerIngo Molnar <mingo@kernel.org>2019-06-25 11:23:47 +0300
commit3c88c692c28746473791276f8b42d2c989d6cbe6 (patch)
tree51cb03ef4cbb611f42ea43c8737be894991b656b /arch/x86/kernel
parentea1ed38dba64b64a245ab8ca1406269d17b99485 (diff)
downloadlinux-3c88c692c28746473791276f8b42d2c989d6cbe6.tar.xz
x86/stackframe/32: Provide consistent pt_regs
Currently pt_regs on x86_32 has an oddity in that kernel regs (!user_mode(regs)) are short two entries (esp/ss). This means that any code trying to use them (typically: regs->sp) needs to jump through some unfortunate hoops. Change the entry code to fix this up and create a full pt_regs frame. This then simplifies various trampolines in ftrace and kprobes, the stack unwinder, ptrace, kdump and kgdb. Much thanks to Josh for help with the cleanups! Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Josh Poimboeuf <jpoimboe@redhat.com> Acked-by: Masami Hiramatsu <mhiramat@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Ingo Molnar <mingo@kernel.org>
Diffstat (limited to 'arch/x86/kernel')
-rw-r--r--arch/x86/kernel/crash.c8
-rw-r--r--arch/x86/kernel/ftrace_32.S75
-rw-r--r--arch/x86/kernel/kgdb.c8
-rw-r--r--arch/x86/kernel/kprobes/common.h4
-rw-r--r--arch/x86/kernel/kprobes/core.c29
-rw-r--r--arch/x86/kernel/kprobes/opt.c20
-rw-r--r--arch/x86/kernel/process_32.c16
-rw-r--r--arch/x86/kernel/ptrace.c29
-rw-r--r--arch/x86/kernel/time.c3
-rw-r--r--arch/x86/kernel/unwind_frame.c32
-rw-r--r--arch/x86/kernel/unwind_orc.c2
11 files changed, 79 insertions, 147 deletions
diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c
index 576b2e1bfc12..84e2d3ddd0eb 100644
--- a/arch/x86/kernel/crash.c
+++ b/arch/x86/kernel/crash.c
@@ -73,14 +73,6 @@ static inline void cpu_crash_vmclear_loaded_vmcss(void)
static void kdump_nmi_callback(int cpu, struct pt_regs *regs)
{
-#ifdef CONFIG_X86_32
- struct pt_regs fixed_regs;
-
- if (!user_mode(regs)) {
- crash_fixup_ss_esp(&fixed_regs, regs);
- regs = &fixed_regs;
- }
-#endif
crash_save_cpu(regs, cpu);
/*
diff --git a/arch/x86/kernel/ftrace_32.S b/arch/x86/kernel/ftrace_32.S
index 0344c3aad4a5..073aab525d80 100644
--- a/arch/x86/kernel/ftrace_32.S
+++ b/arch/x86/kernel/ftrace_32.S
@@ -10,6 +10,7 @@
#include <asm/ftrace.h>
#include <asm/nospec-branch.h>
#include <asm/frame.h>
+#include <asm/asm-offsets.h>
# define function_hook __fentry__
EXPORT_SYMBOL(__fentry__)
@@ -90,26 +91,38 @@ END(ftrace_caller)
ENTRY(ftrace_regs_caller)
/*
- * i386 does not save SS and ESP when coming from kernel.
- * Instead, to get sp, &regs->sp is used (see ptrace.h).
- * Unfortunately, that means eflags must be at the same location
- * as the current return ip is. We move the return ip into the
- * regs->ip location, and move flags into the return ip location.
+ * We're here from an mcount/fentry CALL, and the stack frame looks like:
+ *
+ * <previous context>
+ * RET-IP
+ *
+ * The purpose of this function is to call out in an emulated INT3
+ * environment with a stack frame like:
+ *
+ * <previous context>
+ * gap / RET-IP
+ * gap
+ * gap
+ * gap
+ * pt_regs
+ *
+ * We do _NOT_ restore: ss, flags, cs, gs, fs, es, ds
*/
- pushl $__KERNEL_CS
- pushl 4(%esp) /* Save the return ip */
- pushl $0 /* Load 0 into orig_ax */
+ subl $3*4, %esp # RET-IP + 3 gaps
+ pushl %ss # ss
+ pushl %esp # points at ss
+ addl $5*4, (%esp) # make it point at <previous context>
+ pushfl # flags
+ pushl $__KERNEL_CS # cs
+ pushl 7*4(%esp) # ip <- RET-IP
+ pushl $0 # orig_eax
+
pushl %gs
pushl %fs
pushl %es
pushl %ds
- pushl %eax
-
- /* Get flags and place them into the return ip slot */
- pushf
- popl %eax
- movl %eax, 8*4(%esp)
+ pushl %eax
pushl %ebp
pushl %edi
pushl %esi
@@ -119,24 +132,25 @@ ENTRY(ftrace_regs_caller)
ENCODE_FRAME_POINTER
- movl 12*4(%esp), %eax /* Load ip (1st parameter) */
- subl $MCOUNT_INSN_SIZE, %eax /* Adjust ip */
- movl 15*4(%esp), %edx /* Load parent ip (2nd parameter) */
- movl function_trace_op, %ecx /* Save ftrace_pos in 3rd parameter */
- pushl %esp /* Save pt_regs as 4th parameter */
+ movl PT_EIP(%esp), %eax # 1st argument: IP
+ subl $MCOUNT_INSN_SIZE, %eax
+ movl 21*4(%esp), %edx # 2nd argument: parent ip
+ movl function_trace_op, %ecx # 3rd argument: ftrace_pos
+ pushl %esp # 4th argument: pt_regs
GLOBAL(ftrace_regs_call)
call ftrace_stub
- addl $4, %esp /* Skip pt_regs */
+ addl $4, %esp # skip 4th argument
- /* restore flags */
- push 14*4(%esp)
- popf
+ /* place IP below the new SP */
+ movl PT_OLDESP(%esp), %eax
+ movl PT_EIP(%esp), %ecx
+ movl %ecx, -4(%eax)
- /* Move return ip back to its original location */
- movl 12*4(%esp), %eax
- movl %eax, 14*4(%esp)
+ /* place EAX below that */
+ movl PT_EAX(%esp), %ecx
+ movl %ecx, -8(%eax)
popl %ebx
popl %ecx
@@ -144,14 +158,9 @@ GLOBAL(ftrace_regs_call)
popl %esi
popl %edi
popl %ebp
- popl %eax
- popl %ds
- popl %es
- popl %fs
- popl %gs
- /* use lea to not affect flags */
- lea 3*4(%esp), %esp /* Skip orig_ax, ip and cs */
+ lea -8(%eax), %esp
+ popl %eax
jmp .Lftrace_ret
diff --git a/arch/x86/kernel/kgdb.c b/arch/x86/kernel/kgdb.c
index 6690c5652aeb..23297ea64f5f 100644
--- a/arch/x86/kernel/kgdb.c
+++ b/arch/x86/kernel/kgdb.c
@@ -118,14 +118,6 @@ char *dbg_get_reg(int regno, void *mem, struct pt_regs *regs)
#ifdef CONFIG_X86_32
switch (regno) {
- case GDB_SS:
- if (!user_mode(regs))
- *(unsigned long *)mem = __KERNEL_DS;
- break;
- case GDB_SP:
- if (!user_mode(regs))
- *(unsigned long *)mem = kernel_stack_pointer(regs);
- break;
case GDB_GS:
case GDB_FS:
*(unsigned long *)mem = 0xFFFF;
diff --git a/arch/x86/kernel/kprobes/common.h b/arch/x86/kernel/kprobes/common.h
index 2e25f5f3fa1f..7d3a2e2daf01 100644
--- a/arch/x86/kernel/kprobes/common.h
+++ b/arch/x86/kernel/kprobes/common.h
@@ -72,8 +72,8 @@
" popl %edi\n" \
" popl %ebp\n" \
" popl %eax\n" \
- /* Skip ds, es, fs, gs, orig_ax, and ip. Note: don't pop cs here*/\
- " addl $24, %esp\n"
+ /* Skip ds, es, fs, gs, orig_ax, ip, and cs. */\
+ " addl $7*4, %esp\n"
#endif
/* Ensure if the instruction can be boostable */
diff --git a/arch/x86/kernel/kprobes/core.c b/arch/x86/kernel/kprobes/core.c
index 6afd8061dbae..bd17dbb15d6a 100644
--- a/arch/x86/kernel/kprobes/core.c
+++ b/arch/x86/kernel/kprobes/core.c
@@ -56,7 +56,7 @@
DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL;
DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk);
-#define stack_addr(regs) ((unsigned long *)kernel_stack_pointer(regs))
+#define stack_addr(regs) ((unsigned long *)regs->sp)
#define W(row, b0, b1, b2, b3, b4, b5, b6, b7, b8, b9, ba, bb, bc, bd, be, bf)\
(((b0##UL << 0x0)|(b1##UL << 0x1)|(b2##UL << 0x2)|(b3##UL << 0x3) | \
@@ -718,29 +718,27 @@ asm(
".global kretprobe_trampoline\n"
".type kretprobe_trampoline, @function\n"
"kretprobe_trampoline:\n"
-#ifdef CONFIG_X86_64
/* We don't bother saving the ss register */
+#ifdef CONFIG_X86_64
" pushq %rsp\n"
" pushfq\n"
SAVE_REGS_STRING
" movq %rsp, %rdi\n"
" call trampoline_handler\n"
/* Replace saved sp with true return address. */
- " movq %rax, 152(%rsp)\n"
+ " movq %rax, 19*8(%rsp)\n"
RESTORE_REGS_STRING
" popfq\n"
#else
- " pushf\n"
+ " pushl %esp\n"
+ " pushfl\n"
SAVE_REGS_STRING
" movl %esp, %eax\n"
" call trampoline_handler\n"
- /* Move flags to cs */
- " movl 56(%esp), %edx\n"
- " movl %edx, 52(%esp)\n"
- /* Replace saved flags with true return address. */
- " movl %eax, 56(%esp)\n"
+ /* Replace saved sp with true return address. */
+ " movl %eax, 15*4(%esp)\n"
RESTORE_REGS_STRING
- " popf\n"
+ " popfl\n"
#endif
" ret\n"
".size kretprobe_trampoline, .-kretprobe_trampoline\n"
@@ -781,16 +779,13 @@ __used __visible void *trampoline_handler(struct pt_regs *regs)
INIT_HLIST_HEAD(&empty_rp);
kretprobe_hash_lock(current, &head, &flags);
/* fixup registers */
-#ifdef CONFIG_X86_64
regs->cs = __KERNEL_CS;
- /* On x86-64, we use pt_regs->sp for return address holder. */
- frame_pointer = &regs->sp;
-#else
- regs->cs = __KERNEL_CS | get_kernel_rpl();
+#ifdef CONFIG_X86_32
+ regs->cs |= get_kernel_rpl();
regs->gs = 0;
- /* On x86-32, we use pt_regs->flags for return address holder. */
- frame_pointer = &regs->flags;
#endif
+ /* We use pt_regs->sp for return address holder. */
+ frame_pointer = &regs->sp;
regs->ip = trampoline_address;
regs->orig_ax = ~0UL;
diff --git a/arch/x86/kernel/kprobes/opt.c b/arch/x86/kernel/kprobes/opt.c
index 7c361a24c6df..282b4eb67e30 100644
--- a/arch/x86/kernel/kprobes/opt.c
+++ b/arch/x86/kernel/kprobes/opt.c
@@ -102,14 +102,15 @@ asm (
"optprobe_template_call:\n"
ASM_NOP5
/* Move flags to rsp */
- " movq 144(%rsp), %rdx\n"
- " movq %rdx, 152(%rsp)\n"
+ " movq 18*8(%rsp), %rdx\n"
+ " movq %rdx, 19*8(%rsp)\n"
RESTORE_REGS_STRING
/* Skip flags entry */
" addq $8, %rsp\n"
" popfq\n"
#else /* CONFIG_X86_32 */
- " pushf\n"
+ " pushl %esp\n"
+ " pushfl\n"
SAVE_REGS_STRING
" movl %esp, %edx\n"
".global optprobe_template_val\n"
@@ -118,9 +119,13 @@ asm (
".global optprobe_template_call\n"
"optprobe_template_call:\n"
ASM_NOP5
+ /* Move flags into esp */
+ " movl 14*4(%esp), %edx\n"
+ " movl %edx, 15*4(%esp)\n"
RESTORE_REGS_STRING
- " addl $4, %esp\n" /* skip cs */
- " popf\n"
+ /* Skip flags entry */
+ " addl $4, %esp\n"
+ " popfl\n"
#endif
".global optprobe_template_end\n"
"optprobe_template_end:\n"
@@ -152,10 +157,9 @@ optimized_callback(struct optimized_kprobe *op, struct pt_regs *regs)
} else {
struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
/* Save skipped registers */
-#ifdef CONFIG_X86_64
regs->cs = __KERNEL_CS;
-#else
- regs->cs = __KERNEL_CS | get_kernel_rpl();
+#ifdef CONFIG_X86_32
+ regs->cs |= get_kernel_rpl();
regs->gs = 0;
#endif
regs->ip = (unsigned long)op->kp.addr + INT3_SIZE;
diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c
index 2399e910d109..b8ceec4974fe 100644
--- a/arch/x86/kernel/process_32.c
+++ b/arch/x86/kernel/process_32.c
@@ -62,27 +62,21 @@ void __show_regs(struct pt_regs *regs, enum show_regs_mode mode)
{
unsigned long cr0 = 0L, cr2 = 0L, cr3 = 0L, cr4 = 0L;
unsigned long d0, d1, d2, d3, d6, d7;
- unsigned long sp;
- unsigned short ss, gs;
+ unsigned short gs;
- if (user_mode(regs)) {
- sp = regs->sp;
- ss = regs->ss;
+ if (user_mode(regs))
gs = get_user_gs(regs);
- } else {
- sp = kernel_stack_pointer(regs);
- savesegment(ss, ss);
+ else
savesegment(gs, gs);
- }
show_ip(regs, KERN_DEFAULT);
printk(KERN_DEFAULT "EAX: %08lx EBX: %08lx ECX: %08lx EDX: %08lx\n",
regs->ax, regs->bx, regs->cx, regs->dx);
printk(KERN_DEFAULT "ESI: %08lx EDI: %08lx EBP: %08lx ESP: %08lx\n",
- regs->si, regs->di, regs->bp, sp);
+ regs->si, regs->di, regs->bp, regs->sp);
printk(KERN_DEFAULT "DS: %04x ES: %04x FS: %04x GS: %04x SS: %04x EFLAGS: %08lx\n",
- (u16)regs->ds, (u16)regs->es, (u16)regs->fs, gs, ss, regs->flags);
+ (u16)regs->ds, (u16)regs->es, (u16)regs->fs, gs, regs->ss, regs->flags);
if (mode != SHOW_REGS_ALL)
return;
diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c
index a166c960bc9e..1f92958d8165 100644
--- a/arch/x86/kernel/ptrace.c
+++ b/arch/x86/kernel/ptrace.c
@@ -154,35 +154,6 @@ static inline bool invalid_selector(u16 value)
#define FLAG_MASK FLAG_MASK_32
-/*
- * X86_32 CPUs don't save ss and esp if the CPU is already in kernel mode
- * when it traps. The previous stack will be directly underneath the saved
- * registers, and 'sp/ss' won't even have been saved. Thus the '&regs->sp'.
- *
- * Now, if the stack is empty, '&regs->sp' is out of range. In this
- * case we try to take the previous stack. To always return a non-null
- * stack pointer we fall back to regs as stack if no previous stack
- * exists.
- *
- * This is valid only for kernel mode traps.
- */
-unsigned long kernel_stack_pointer(struct pt_regs *regs)
-{
- unsigned long context = (unsigned long)regs & ~(THREAD_SIZE - 1);
- unsigned long sp = (unsigned long)&regs->sp;
- u32 *prev_esp;
-
- if (context == (sp & ~(THREAD_SIZE - 1)))
- return sp;
-
- prev_esp = (u32 *)(context);
- if (*prev_esp)
- return (unsigned long)*prev_esp;
-
- return (unsigned long)regs;
-}
-EXPORT_SYMBOL_GPL(kernel_stack_pointer);
-
static unsigned long *pt_regs_access(struct pt_regs *regs, unsigned long regno)
{
BUILD_BUG_ON(offsetof(struct pt_regs, bx) != 0);
diff --git a/arch/x86/kernel/time.c b/arch/x86/kernel/time.c
index 0e14f6c0d35e..2acbab329754 100644
--- a/arch/x86/kernel/time.c
+++ b/arch/x86/kernel/time.c
@@ -37,8 +37,7 @@ unsigned long profile_pc(struct pt_regs *regs)
#ifdef CONFIG_FRAME_POINTER
return *(unsigned long *)(regs->bp + sizeof(long));
#else
- unsigned long *sp =
- (unsigned long *)kernel_stack_pointer(regs);
+ unsigned long *sp = (unsigned long *)regs->sp;
/*
* Return address is either directly at stack pointer
* or above a saved flags. Eflags has bits 22-31 zero,
diff --git a/arch/x86/kernel/unwind_frame.c b/arch/x86/kernel/unwind_frame.c
index 6106760de716..a224b5ab103f 100644
--- a/arch/x86/kernel/unwind_frame.c
+++ b/arch/x86/kernel/unwind_frame.c
@@ -70,15 +70,6 @@ static void unwind_dump(struct unwind_state *state)
}
}
-static size_t regs_size(struct pt_regs *regs)
-{
- /* x86_32 regs from kernel mode are two words shorter: */
- if (IS_ENABLED(CONFIG_X86_32) && !user_mode(regs))
- return sizeof(*regs) - 2*sizeof(long);
-
- return sizeof(*regs);
-}
-
static bool in_entry_code(unsigned long ip)
{
char *addr = (char *)ip;
@@ -198,12 +189,6 @@ static struct pt_regs *decode_frame_pointer(unsigned long *bp)
}
#endif
-#ifdef CONFIG_X86_32
-#define KERNEL_REGS_SIZE (sizeof(struct pt_regs) - 2*sizeof(long))
-#else
-#define KERNEL_REGS_SIZE (sizeof(struct pt_regs))
-#endif
-
static bool update_stack_state(struct unwind_state *state,
unsigned long *next_bp)
{
@@ -214,7 +199,7 @@ static bool update_stack_state(struct unwind_state *state,
size_t len;
if (state->regs)
- prev_frame_end = (void *)state->regs + regs_size(state->regs);
+ prev_frame_end = (void *)state->regs + sizeof(*state->regs);
else
prev_frame_end = (void *)state->bp + FRAME_HEADER_SIZE;
@@ -222,7 +207,7 @@ static bool update_stack_state(struct unwind_state *state,
regs = decode_frame_pointer(next_bp);
if (regs) {
frame = (unsigned long *)regs;
- len = KERNEL_REGS_SIZE;
+ len = sizeof(*regs);
state->got_irq = true;
} else {
frame = next_bp;
@@ -246,14 +231,6 @@ static bool update_stack_state(struct unwind_state *state,
frame < prev_frame_end)
return false;
- /*
- * On 32-bit with user mode regs, make sure the last two regs are safe
- * to access:
- */
- if (IS_ENABLED(CONFIG_X86_32) && regs && user_mode(regs) &&
- !on_stack(info, frame, len + 2*sizeof(long)))
- return false;
-
/* Move state to the next frame: */
if (regs) {
state->regs = regs;
@@ -412,10 +389,9 @@ void __unwind_start(struct unwind_state *state, struct task_struct *task,
* Pretend that the frame is complete and that BP points to it, but save
* the real BP so that we can use it when looking for the next frame.
*/
- if (regs && regs->ip == 0 &&
- (unsigned long *)kernel_stack_pointer(regs) >= first_frame) {
+ if (regs && regs->ip == 0 && (unsigned long *)regs->sp >= first_frame) {
state->next_bp = bp;
- bp = ((unsigned long *)kernel_stack_pointer(regs)) - 1;
+ bp = ((unsigned long *)regs->sp) - 1;
}
/* Initialize stack info and make sure the frame data is accessible: */
diff --git a/arch/x86/kernel/unwind_orc.c b/arch/x86/kernel/unwind_orc.c
index 33b66b5c5aec..7f79aeebe3ee 100644
--- a/arch/x86/kernel/unwind_orc.c
+++ b/arch/x86/kernel/unwind_orc.c
@@ -580,7 +580,7 @@ void __unwind_start(struct unwind_state *state, struct task_struct *task,
goto done;
state->ip = regs->ip;
- state->sp = kernel_stack_pointer(regs);
+ state->sp = regs->sp;
state->bp = regs->bp;
state->regs = regs;
state->full_regs = true;