Commit f1d5df84 authored by Ilya Leoshkevich's avatar Ilya Leoshkevich Committed by Alexei Starovoitov

s390/bpf: Implement bpf_arch_text_poke()

bpf_arch_text_poke() is used to hotpatch eBPF programs and trampolines.
s390x has a very strict hotpatching restriction: the only thing that is
allowed to be hotpatched is conditional branch mask.

Take the same approach as commit de5012b4 ("s390/ftrace: implement
hotpatching"): create a conditional jump to a "plt", which loads the
target address from memory and jumps to it; then first patch this
address, and then the mask.

Trampolines (introduced in the next patch) respect the ftrace calling
convention: the return address is in %r0, and %r1 is clobbered. With
that in mind, bpf_arch_text_poke() does not differentiate between jumps
and calls.

However, there is a simple optimization for jumps (for the epilogue_ip
case): if a jump already points to the destination, then there is no
"plt" and we can just flip the mask.

For simplicity, the "plt" template is defined in assembly, and its size
is used to define C arrays. There doesn't seem to be a way to convey
this size to C as a constant, so it's hardcoded and double-checked
during runtime.
Signed-off-by: default avatarIlya Leoshkevich <iii@linux.ibm.com>
Link: https://lore.kernel.org/r/20230129190501.1624747-4-iii@linux.ibm.comSigned-off-by: default avatarAlexei Starovoitov <ast@kernel.org>
parent bb4ef8fc
......@@ -30,6 +30,7 @@
#include <asm/facility.h>
#include <asm/nospec-branch.h>
#include <asm/set_memory.h>
#include <asm/text-patching.h>
#include "bpf_jit.h"
struct bpf_jit {
......@@ -50,6 +51,8 @@ struct bpf_jit {
int r14_thunk_ip; /* Address of expoline thunk for 'br %r14' */
int tail_call_start; /* Tail call start offset */
int excnt; /* Number of exception table entries */
int prologue_plt_ret; /* Return address for prologue hotpatch PLT */
int prologue_plt; /* Start of prologue hotpatch PLT */
};
#define SEEN_MEM BIT(0) /* use mem[] for temporary storage */
......@@ -506,6 +509,36 @@ static void bpf_skip(struct bpf_jit *jit, int size)
}
}
/*
* PLT for hotpatchable calls. The calling convention is the same as for the
* ftrace hotpatch trampolines: %r0 is return address, %r1 is clobbered.
*/
extern const char bpf_plt[];
extern const char bpf_plt_ret[];
extern const char bpf_plt_target[];
extern const char bpf_plt_end[];
#define BPF_PLT_SIZE 32
asm(
".pushsection .rodata\n"
" .align 8\n"
"bpf_plt:\n"
" lgrl %r0,bpf_plt_ret\n"
" lgrl %r1,bpf_plt_target\n"
" br %r1\n"
" .align 8\n"
"bpf_plt_ret: .quad 0\n"
"bpf_plt_target: .quad 0\n"
"bpf_plt_end:\n"
" .popsection\n"
);
static void bpf_jit_plt(void *plt, void *ret, void *target)
{
memcpy(plt, bpf_plt, BPF_PLT_SIZE);
*(void **)((char *)plt + (bpf_plt_ret - bpf_plt)) = ret;
*(void **)((char *)plt + (bpf_plt_target - bpf_plt)) = target;
}
/*
* Emit function prologue
*
......@@ -514,6 +547,11 @@ static void bpf_skip(struct bpf_jit *jit, int size)
*/
static void bpf_jit_prologue(struct bpf_jit *jit, u32 stack_depth)
{
/* No-op for hotpatching */
/* brcl 0,prologue_plt */
EMIT6_PCREL_RILC(0xc0040000, 0, jit->prologue_plt);
jit->prologue_plt_ret = jit->prg;
if (jit->seen & SEEN_TAIL_CALL) {
/* xc STK_OFF_TCCNT(4,%r15),STK_OFF_TCCNT(%r15) */
_EMIT6(0xd703f000 | STK_OFF_TCCNT, 0xf000 | STK_OFF_TCCNT);
......@@ -589,6 +627,13 @@ static void bpf_jit_epilogue(struct bpf_jit *jit, u32 stack_depth)
/* br %r1 */
_EMIT2(0x07f1);
}
jit->prg = ALIGN(jit->prg, 8);
jit->prologue_plt = jit->prg;
if (jit->prg_buf)
bpf_jit_plt(jit->prg_buf + jit->prg,
jit->prg_buf + jit->prologue_plt_ret, NULL);
jit->prg += BPF_PLT_SIZE;
}
static int get_probe_mem_regno(const u8 *insn)
......@@ -1776,6 +1821,9 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
struct bpf_jit jit;
int pass;
if (WARN_ON_ONCE(bpf_plt_end - bpf_plt != BPF_PLT_SIZE))
return orig_fp;
if (!fp->jit_requested)
return orig_fp;
......@@ -1867,3 +1915,52 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
tmp : orig_fp);
return fp;
}
int bpf_arch_text_poke(void *ip, enum bpf_text_poke_type t,
void *old_addr, void *new_addr)
{
struct {
u16 opc;
s32 disp;
} __packed insn;
char expected_plt[BPF_PLT_SIZE];
char current_plt[BPF_PLT_SIZE];
char *plt;
int err;
/* Verify the branch to be patched. */
err = copy_from_kernel_nofault(&insn, ip, sizeof(insn));
if (err < 0)
return err;
if (insn.opc != (0xc004 | (old_addr ? 0xf0 : 0)))
return -EINVAL;
if (t == BPF_MOD_JUMP &&
insn.disp == ((char *)new_addr - (char *)ip) >> 1) {
/*
* The branch already points to the destination,
* there is no PLT.
*/
} else {
/* Verify the PLT. */
plt = (char *)ip + (insn.disp << 1);
err = copy_from_kernel_nofault(current_plt, plt, BPF_PLT_SIZE);
if (err < 0)
return err;
bpf_jit_plt(expected_plt, (char *)ip + 6, old_addr);
if (memcmp(current_plt, expected_plt, BPF_PLT_SIZE))
return -EINVAL;
/* Adjust the call address. */
s390_kernel_write(plt + (bpf_plt_target - bpf_plt),
&new_addr, sizeof(void *));
}
/* Adjust the mask of the branch. */
insn.opc = 0xc004 | (new_addr ? 0xf0 : 0);
s390_kernel_write((char *)ip + 1, (char *)&insn.opc + 1, 1);
/* Make the new code visible to the other CPUs. */
text_poke_sync_lock();
return 0;
}
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment