ruby/zjit.c

#include "internal.h"
#include "internal/sanitizers.h"
#include "internal/string.h"
#include "internal/hash.h"
#include "internal/variable.h"
#include "internal/compile.h"
#include "internal/class.h"
#include "internal/fixnum.h"
#include "internal/numeric.h"
#include "internal/gc.h"
#include "internal/vm.h"
#include "vm_core.h"
#include "vm_callinfo.h"
#include "builtin.h"
#include "insns.inc"
#include "insns_info.inc"
#include "vm_sync.h"
#include "vm_insnhelper.h"
#include "probes.h"
#include "probes_helper.h"
#include "iseq.h"
#include "ruby/debug.h"
#include "internal/cont.h"
#include "zjit.h"

// For mmapp(), sysconf()
#ifndef _WIN32
#include <unistd.h>
#include <sys/mman.h>
#endif

#include <errno.h>

uint32_t
rb_zjit_get_page_size(void)
{
#if defined(_SC_PAGESIZE)
    long page_size = sysconf(_SC_PAGESIZE);
    if (page_size <= 0) rb_bug("zjit: failed to get page size");

    // 1 GiB limit. x86 CPUs with PDPE1GB can do this and anything larger is unexpected.
    // Though our design sort of assume we have fine grained control over memory protection
    // which require small page sizes.
    if (page_size > 0x40000000l) rb_bug("zjit page size too large");

    return (uint32_t)page_size;
#else
#error "ZJIT supports POSIX only for now"
#endif
}

#if defined(MAP_FIXED_NOREPLACE) && defined(_SC_PAGESIZE)
// Align the current write position to a multiple of bytes
static uint8_t *
align_ptr(uint8_t *ptr, uint32_t multiple)
{
    // Compute the pointer modulo the given alignment boundary
    uint32_t rem = ((uint32_t)(uintptr_t)ptr) % multiple;

    // If the pointer is already aligned, stop
    if (rem == 0)
        return ptr;

    // Pad the pointer by the necessary amount to align it
    uint32_t pad = multiple - rem;

    return ptr + pad;
}
#endif

// Address space reservation. Memory pages are mapped on an as needed basis.
// See the Rust mm module for details.
uint8_t *
rb_zjit_reserve_addr_space(uint32_t mem_size)
{
#ifndef _WIN32
    uint8_t *mem_block;

    // On Linux
    #if defined(MAP_FIXED_NOREPLACE) && defined(_SC_PAGESIZE)
        uint32_t const page_size = (uint32_t)sysconf(_SC_PAGESIZE);
        uint8_t *const cfunc_sample_addr = (void *)(uintptr_t)&rb_zjit_reserve_addr_space;
        uint8_t *const probe_region_end = cfunc_sample_addr + INT32_MAX;
        // Align the requested address to page size
        uint8_t *req_addr = align_ptr(cfunc_sample_addr, page_size);

        // Probe for addresses close to this function using MAP_FIXED_NOREPLACE
        // to improve odds of being in range for 32-bit relative call instructions.
        do {
            mem_block = mmap(
                req_addr,
                mem_size,
                PROT_NONE,
                MAP_PRIVATE | MAP_ANONYMOUS | MAP_FIXED_NOREPLACE,
                -1,
                0
            );

            // If we succeeded, stop
            if (mem_block != MAP_FAILED) {
                ruby_annotate_mmap(mem_block, mem_size, "Ruby:rb_zjit_reserve_addr_space");
                break;
            }

            // -4MiB. Downwards to probe away from the heap. (On x86/A64 Linux
            // main_code_addr < heap_addr, and in case we are in a shared
            // library mapped higher than the heap, downwards is still better
            // since it's towards the end of the heap rather than the stack.)
            req_addr -= 4 * 1024 * 1024;
        } while (req_addr < probe_region_end);

    // On MacOS and other platforms
    #else
        // Try to map a chunk of memory as executable
        mem_block = mmap(
            (void *)rb_zjit_reserve_addr_space,
            mem_size,
            PROT_NONE,
            MAP_PRIVATE | MAP_ANONYMOUS,
            -1,
            0
        );
    #endif

    // Fallback
    if (mem_block == MAP_FAILED) {
        // Try again without the address hint (e.g., valgrind)
        mem_block = mmap(
            NULL,
            mem_size,
            PROT_NONE,
            MAP_PRIVATE | MAP_ANONYMOUS,
            -1,
            0
        );

        if (mem_block != MAP_FAILED) {
            ruby_annotate_mmap(mem_block, mem_size, "Ruby:rb_zjit_reserve_addr_space:fallback");
        }
    }

    // Check that the memory mapping was successful
    if (mem_block == MAP_FAILED) {
        perror("ruby: zjit: mmap:");
        if(errno == ENOMEM) {
            // No crash report if it's only insufficient memory
            exit(EXIT_FAILURE);
        }
        rb_bug("mmap failed");
    }

    return mem_block;
#else
    // Windows not supported for now
    return NULL;
#endif
}

void rb_zjit_profile_disable(const rb_iseq_t *iseq);

void
rb_zjit_compile_iseq(const rb_iseq_t *iseq, rb_execution_context_t *ec, bool jit_exception)
{
    RB_VM_LOCK_ENTER();
    rb_vm_barrier();

    // Convert ZJIT instructions back to bare instructions
    rb_zjit_profile_disable(iseq);

    // Compile a block version starting at the current instruction
    uint8_t *rb_zjit_iseq_gen_entry_point(const rb_iseq_t *iseq, rb_execution_context_t *ec); // defined in Rust
    uintptr_t code_ptr = (uintptr_t)rb_zjit_iseq_gen_entry_point(iseq, ec);

    // TODO: support jit_exception
    iseq->body->jit_entry = (rb_jit_func_t)code_ptr;

    RB_VM_LOCK_LEAVE();
}

extern VALUE *rb_vm_base_ptr(struct rb_control_frame_struct *cfp);

bool
rb_zjit_multi_ractor_p(void)
{
    return rb_multi_ractor_p();
}

bool
rb_zjit_constcache_shareable(const struct iseq_inline_constant_cache_entry *ice)
{
    return (ice->flags & IMEMO_CONST_CACHE_SHAREABLE) != 0;
}

// Release the VM lock. The lock level must point to the same integer used to
// acquire the lock.
void
rb_zjit_vm_unlock(unsigned int *recursive_lock_level, const char *file, int line)
{
    rb_vm_lock_leave(recursive_lock_level, file, line);
}

bool
rb_zjit_mark_writable(void *mem_block, uint32_t mem_size)
{
    return mprotect(mem_block, mem_size, PROT_READ | PROT_WRITE) == 0;
}

void
rb_zjit_mark_executable(void *mem_block, uint32_t mem_size)
{
    // Do not call mprotect when mem_size is zero. Some platforms may return
    // an error for it. https://github.com/Shopify/ruby/issues/450
    if (mem_size == 0) {
        return;
    }
    if (mprotect(mem_block, mem_size, PROT_READ | PROT_EXEC)) {
        rb_bug("Couldn't make JIT page (%p, %lu bytes) executable, errno: %s",
            mem_block, (unsigned long)mem_size, strerror(errno));
    }
}

// Free the specified memory block.
bool
rb_zjit_mark_unused(void *mem_block, uint32_t mem_size)
{
    // On Linux, you need to use madvise MADV_DONTNEED to free memory.
    // We might not need to call this on macOS, but it's not really documented.
    // We generally prefer to do the same thing on both to ease testing too.
    madvise(mem_block, mem_size, MADV_DONTNEED);

    // On macOS, mprotect PROT_NONE seems to reduce RSS.
    // We also call this on Linux to avoid executing unused pages.
    return mprotect(mem_block, mem_size, PROT_NONE) == 0;
}

// Invalidate icache for arm64.
// `start` is inclusive and `end` is exclusive.
void
rb_zjit_icache_invalidate(void *start, void *end)
{
    // Clear/invalidate the instruction cache. Compiles to nothing on x86_64
    // but required on ARM before running freshly written code.
    // On Darwin it's the same as calling sys_icache_invalidate().
#ifdef __GNUC__
    __builtin___clear_cache(start, end);
#elif defined(__aarch64__)
#error No instruction cache clear available with this compiler on Aarch64!
#endif
}

// Acquire the VM lock and then signal all other Ruby threads (ractors) to
// contend for the VM lock, putting them to sleep. ZJIT uses this to evict
// threads running inside generated code so among other things, it can
// safely change memory protection of regions housing generated code.
void
rb_zjit_vm_lock_then_barrier(unsigned int *recursive_lock_level, const char *file, int line)
{
    rb_vm_lock_enter(recursive_lock_level, file, line);
    rb_vm_barrier();
}

// Convert a given ISEQ's instructions to zjit_* instructions
void
rb_zjit_profile_enable(const rb_iseq_t *iseq)
{
    // This table encodes an opcode into the instruction's address
    const void *const *insn_table = rb_vm_get_insns_address_table();

    unsigned int insn_idx = 0;
    while (insn_idx < iseq->body->iseq_size) {
        int insn = rb_vm_insn_addr2opcode((void *)iseq->body->iseq_encoded[insn_idx]);
        int zjit_insn = vm_bare_insn_to_zjit_insn(insn);
        if (insn != zjit_insn) {
            iseq->body->iseq_encoded[insn_idx] = (VALUE)insn_table[zjit_insn];
        }
        insn_idx += insn_len(insn);
    }
}

// Convert a given ISEQ's ZJIT instructions to bare instructions
void
rb_zjit_profile_disable(const rb_iseq_t *iseq)
{
    // This table encodes an opcode into the instruction's address
    const void *const *insn_table = rb_vm_get_insns_address_table();

    unsigned int insn_idx = 0;
    while (insn_idx < iseq->body->iseq_size) {
        int insn = rb_vm_insn_addr2opcode((void *)iseq->body->iseq_encoded[insn_idx]);
        int bare_insn = vm_zjit_insn_to_bare_insn(insn);
        if (insn != bare_insn) {
            iseq->body->iseq_encoded[insn_idx] = (VALUE)insn_table[bare_insn];
        }
        insn_idx += insn_len(insn);
    }
}

// Get profiling information for ISEQ
void *
rb_iseq_get_zjit_payload(const rb_iseq_t *iseq)
{
    RUBY_ASSERT_ALWAYS(IMEMO_TYPE_P(iseq, imemo_iseq));
    if (iseq->body) {
        return iseq->body->zjit_payload;
    }
    else {
        // Body is NULL when constructing the iseq.
        return NULL;
    }
}

// Set profiling information for ISEQ
void
rb_iseq_set_zjit_payload(const rb_iseq_t *iseq, void *payload)
{
    RUBY_ASSERT_ALWAYS(IMEMO_TYPE_P(iseq, imemo_iseq));
    RUBY_ASSERT_ALWAYS(iseq->body);
    RUBY_ASSERT_ALWAYS(NULL == iseq->body->zjit_payload);
    iseq->body->zjit_payload = payload;
}

// Primitives used by zjit.rb
VALUE rb_zjit_assert_compiles(rb_execution_context_t *ec, VALUE self);

void
rb_zjit_print_exception(void)
{
    VALUE exception = rb_errinfo();
    rb_set_errinfo(Qnil);
    assert(RTEST(exception));
    rb_warn("Ruby error: %"PRIsVALUE"", rb_funcall(exception, rb_intern("full_message"), 0));
}

// Preprocessed zjit.rb generated during build
#include "zjit.rbinc"