LithosAnanake/v4/system/boot.c

/* boot.c -- the nucleus, then its capsules, then the prompt.  See boot.h.
 *
 * Nothing here uses the C library: the bare-metal kernel links this file.
 * It is not part of the engine (v4/src): it needs the capsule directory,
 * which only a whole system has.
 */
#include "v4/blocks.h"
#include "v4/boot.h"
#include "v4/message.h"
#include "starkernel/capsule.h"
#include "starkernel/capsule_generated.h"
#include "starkernel/capsule_sig.h"
#include "starkernel/capsule_blocks.h"

/* The capsules, in the order they are loaded. */
/* POST is part of the boot: v4:post79.4th is loaded after the vocabulary it
 * tests and must end with a clean tally.  -DV4_POST_AT_BOOT=0 leaves it
 * out, for work on a vocabulary that does not pass yet. */
#ifndef V4_POST_AT_BOOT
#define V4_POST_AT_BOOT 1
#endif
static const char *const boot_capsules[] = {
    "v4:forth79.4th",
#if V4_POST_AT_BOOT
    "v4:post79.4th",
#endif
};

#define LINE_MAX   1025u                 /* a line is at most 1024 characters, a block */
#define STEP_LIMIT 4000000000ULL         /* instruction words one line may take */

/* ---- printing ------------------------------------------------------------ */

static void say(const v4_boot *b, const char *s)
{
    unsigned len = 0;
    while (s[len]) len++;
    b->out(s, len);
}

static void say_dec(const v4_boot *b, uint32_t v)
{
    char buf[10];
    unsigned i = sizeof buf;
    do { buf[--i] = (char)('0' + v % 10u); v /= 10u; } while (v);
    b->out(buf + i, (unsigned)sizeof buf - i);
}

static void say_hex(const v4_boot *b, uint64_t v)
{
    char buf[18];
    unsigned i;
    buf[0] = '0'; buf[1] = 'x';
    for (i = 0; i < 16; i++) buf[2 + i] = "0123456789abcdef"[(v >> (60 - 4 * i)) & 15u];
    b->out(buf, sizeof buf);
}

/* ---- hashing: FNV-1a, 64 bits, as kernel/src/vm/parity.c ----------------- */

uint64_t v4_boot_dict_hash(const v4_node *n, const v4_image *im)
{
    return v4_image_dict_hash(n, im);
}

/* how many words FORTH holds: the list from LATEST, each entry's link in the
 * cell before its code */
static uint32_t word_count(const v4_node *n, const v4_image *im)
{
    v4_cell xt = n->mem[im->latest];
    uint32_t count = 0;
    while (xt > 0 && xt < (v4_cell)V4_NODE_WORDS && count < (uint32_t)V4_NODE_WORDS) {
        count++;
        xt = n->mem[xt - 1];
    }
    return count;
}

/* ---- handing the node a line ---------------------------------------------- */

/* POST's verdict.  The POST capsule ends by printing one line,
 *     PARITY:V4_POST tests=N pass=N fail=N
 * and the boot passes only if it has seen that line with fail=0.  That no
 * line was refused is not enough: a POST that was broken half way would
 * refuse nothing. */
static char     post_line[96];
static unsigned post_len;
static int      post_seen, post_clean;

static int text_at(const char *line, unsigned len, unsigned at, const char *want)
{
    unsigned i;
    for (i = 0; want[i]; i++)
        if (at + i >= len || line[at + i] != want[i]) return 0;
    return 1;
}

static void post_watch(char c)
{
    static const char clean[] = " fail=0";
    if (c != '\n') {
        if (post_len < sizeof post_line) post_line[post_len++] = c;
        return;
    }
    if (text_at(post_line, post_len, 0, "PARITY:V4_POST ")) {
        post_seen = 1;
        post_clean = post_len >= sizeof clean - 1 && text_at(post_line, post_len, post_len - (unsigned)(sizeof clean - 1), clean);
    }
    post_len = 0;
}

/* The boot is the node's neighbour on two of its ports.  On port 0 it is
 * the kernel: the node writes the number of a request there (ENGINE.md
 * 3.3) -- for a block (blocks.h), or for one of the host's words.  On port
 * 1 it is the console: it writes the node a message of text and reads back
 * messages of what the node printed and of how the text ended (message.h). */
#define KERNEL_PORT  0u
#define CONSOLE_PORT 1u
#define CONSOLE_ID   1                   /* the console's number as a sender; the node's is 0 until it is given one */

int v4_boot_line(const v4_boot *b, const char *text, unsigned len)
{
    static v4_message out, in;           /* one line at a time: the boot is not re-entered */
    v4_node *n = b->n;
    uint64_t steps = 0;
    unsigned sent = 0, i;

    if (!v4_message_text(&out, 0, CONSOLE_ID, V4_MSG_TEXT, text, len)) return V4_BOOT_LINE_TOO_LONG;
    in.count = 0;
    for (;;) {
        (void)v4_exec_step_word(n, b->es, b->h);
        if (n->stopped) return V4_BOOT_LINE_STOPPED;

        if (n->asking && n->ask_port == KERNEL_PORT) {         /* a request: the kernel's turn */
            if (v4_blocks_serve(n, n->request)) {              /* a block, as for any node */
                v4_node_port_served(n);
            } else if (b->serve && n->request >= 1 && n->request <= (v4_cell)b->word_count) {
                b->serve(n, (unsigned)n->request);
                v4_node_port_served(n);
            } else {
                /* a request no one serves is an error, as any other (D-18):
                 * 12, Argument out of range */
                v4_node_port_served(n);
                v4_node_store(n, n->error_reg, 12);      /* a store to NODE-ERROR raises it */
            }
            continue;
        }
        if (n->asking && n->ask_port == CONSOLE_PORT) {        /* a word of a message from the node */
            v4_cell word = n->request;
            v4_node_port_served(n);
            if (v4_message_word(&in, word)) {
                unsigned chars = v4_message_length(&in);
                if (v4_message_type(&in) == V4_MSG_OUTPUT) {
                    for (i = 0; i < chars; i++) {
                        char c = v4_message_char(&in, i);
                        post_watch(c);
                        b->out(&c, 1);
                    }
                } else if (v4_message_type(&in) == V4_MSG_DONE) {
                    return (int)in.word[V4_MSG_HEADER];
                }
                in.count = 0;
            }
            continue;
        }
        if (n->asking) return V4_BOOT_LINE_STOPPED;            /* a port with nothing on it */

        if (n->reading && !n->given) {
            if (sent < out.count && (n->read_port == CONSOLE_PORT || n->read_port == V4_PORT_ANY)) {
                v4_node_port_give(n, CONSOLE_PORT, out.word[sent++]);   /* a word of the message to the node */
                continue;
            }
            return V4_BOOT_LINE_STOPPED;                       /* waiting for what will not come */
        }
        if (v4_image_waiting(n, b->im)) {             /* the text is reading the keyboard */
            int c = b->key ? b->key() : V4_BOOT_KEY_END;
            unsigned char ch = (unsigned char)c;
            if (c == V4_BOOT_KEY_END) return V4_BOOT_LINE_NO_INPUT;
            if (c >= 0) (void)v4_node_console_feed(n, &ch, 1u);
            continue;
        }
        if (++steps > STEP_LIMIT) return V4_BOOT_LINE_STOPPED;
    }
}

/* ---- one capsule --------------------------------------------------------- */

static void say_where(const v4_boot *b, const char *name, uint32_t block, uint32_t line)
{
    say(b, "\nV4: capsule "); say(b, name);
    say(b, " block "); say_dec(b, block);
    say(b, " line "); say_dec(b, line);
}

static int load_capsule(const v4_boot *b, const char *name)
{
    const CapsuleDirHeader *dir   = capsule_get_directory();
    const CapsuleDesc      *descs = capsule_get_descriptors();
    const CapsuleNameEntry *names = capsule_get_names();
    const uint8_t          *arena = capsule_get_arena();
    const CapsuleDesc      *cap;
    const uint8_t          *p, *end;
    CapsuleValidateResult   vr;
    CapsuleSigResult        sr;
    uint32_t                block = 0, line = 0;
    int                     in_block = 0;

    cap = capsule_find_by_name(dir, descs, names, name);
    if (!cap) { say(b, "V4: capsule "); say(b, name); say(b, " is not in this binary\n"); return 0; }

    vr = capsule_validate(cap, arena, dir->arena_size, 1);
    if (vr != CAPSULE_VALID) {
        say(b, "V4: capsule "); say(b, name); say(b, ": "); say(b, capsule_validate_result_str(vr)); say(b, "\n");
        return 0;
    }

    sr = capsule_verify_signature(descs, names, capsule_get_signatures(), arena, dir->desc_count, (int)(cap - descs));
    if (sr != CAPSULE_SIG_OK) {
        say(b, "V4: capsule "); say(b, name); say(b, " signature: "); say(b, capsule_sig_result_str(sr)); say(b, "\n");
        if (sr == CAPSULE_SIG_INVALID) return 0;
    }

    p = capsule_get_payload(cap, arena);
    if (!p) { say(b, "V4: capsule "); say(b, name); say(b, " has no payload\n"); return 0; }
    end = p + cap->length;

    while (p < end) {
        const uint8_t *after, *nl;
        uint32_t num;
        unsigned len, i;
        char text[LINE_MAX];

        if (capsule_block_header(p, end, &num, &after)) {
            block = num; line = 0; in_block = 1; p = after;
            continue;
        }
        for (nl = p; nl < end && *nl != '\n'; nl++) { }
        len = (unsigned)(nl - p);
        if (len && p[len - 1] == '\r') len--;
        if (in_block) {
            line++;
            if (len >= LINE_MAX) { say_where(b, name, block, line); say(b, " is too long\n"); return 0; }
            if (len) {
                for (i = 0; i < len; i++) text[i] = (char)p[i];
                int how = v4_boot_line(b, text, len);
                if (how != V4_TEXT_COMPLETED) {
                    say_where(b, name, block, line);
                    say(b, how == V4_TEXT_ERROR ? " ended in an error: "
                         : how == V4_TEXT_QUIT ? " ended with QUIT: "
                         : how == V4_BOOT_LINE_NO_INPUT ? " reads the keyboard: "
                         : " stopped the node: ");
                    b->out(text, len); say(b, "\n");
                    return 0;
                }
            }
        }
        p = (nl < end) ? nl + 1 : end;
    }

    say(b, "PARITY:V4_CAPSULE name="); say(b, name);
    say(b, " capsule_id="); say_hex(b, cap->capsule_id);
    say(b, " capsule_hash="); say_hex(b, cap->content_hash);
    say(b, " dict_hash="); say_hex(b, v4_boot_dict_hash(b->n, b->im));
    say(b, "\n");
    return 1;
}

/* ---- the nucleus ----------------------------------------------------------- */

/* The nucleus is a capsule of F18 code (v4/include/v4/capsule.h): the words
 * themselves, to be written to the port of a node that has nothing in it.
 * The boot is that node's neighbour.  It finds the capsule, checks its hash
 * and its signature as it does any capsule's, and gives the node a word
 * each time it reads its port, until the node has taken them all and is
 * where the nucleus waits to be handed a line. */
#define NUCLEUS_NAME_(bits) "v4:nucleus-" #bits ".f18"
#define NUCLEUS_NAME__(bits) NUCLEUS_NAME_(bits)
#define NUCLEUS_NAME NUCLEUS_NAME__(V4_CELL_BITS)
#define CELL_BYTES (V4_CELL_BITS / 8)

static int load_nucleus(const v4_boot *b)
{
    const CapsuleDirHeader *dir   = capsule_get_directory();
    const CapsuleDesc      *descs = capsule_get_descriptors();
    const CapsuleNameEntry *names = capsule_get_names();
    const uint8_t          *arena = capsule_get_arena();
    const CapsuleDesc      *cap;
    const uint8_t          *p;
    CapsuleValidateResult   vr;
    CapsuleSigResult        sr;
    v4_node                *n = b->n;
    uint64_t                words, at = 0, steps = 0;

    cap = capsule_find_by_name(dir, descs, names, NUCLEUS_NAME);
    if (!cap) { say(b, "V4: capsule " NUCLEUS_NAME " is not in this binary\n"); return 0; }
    vr = capsule_validate(cap, arena, dir->arena_size, 1);
    if (vr != CAPSULE_VALID) {
        say(b, "V4: capsule " NUCLEUS_NAME ": "); say(b, capsule_validate_result_str(vr)); say(b, "\n");
        return 0;
    }
    sr = capsule_verify_signature(descs, names, capsule_get_signatures(), arena, dir->desc_count, (int)(cap - descs));
    if (sr != CAPSULE_SIG_OK) {
        say(b, "V4: capsule " NUCLEUS_NAME " signature: "); say(b, capsule_sig_result_str(sr)); say(b, "\n");
        if (sr == CAPSULE_SIG_INVALID) return 0;
    }
    p = capsule_get_payload(cap, arena);
    if (!p || cap->length % CELL_BYTES != 0) { say(b, "V4: capsule " NUCLEUS_NAME " is not whole words\n"); return 0; }
    words = cap->length / CELL_BYTES;

    /* until it has taken every word and is waiting for a message: reading
     * its ports again, but from its memory now and not from the port */
    while (at < words || !(n->reading && !n->given && n->p != b->im->port + (v4_cell)V4_PORT_ANY)) {
        (void)v4_exec_step_word(n, b->es, b->h);
        if (n->stopped || n->asking || ++steps > STEP_LIMIT) {
            say(b, "V4: the node did not take the nucleus in\n");
            return 0;
        }
        if (n->reading && !n->given) {
            v4_ucell word = 0;
            unsigned i;
            if (at >= words) {
                if (n->p != b->im->port + (v4_cell)V4_PORT_ANY) break;      /* it is waiting for a message */
                say(b, "V4: the node wants more than the nucleus holds\n");
                return 0;
            }
            for (i = 0; i < CELL_BYTES; i++) word |= (v4_ucell)p[at * CELL_BYTES + i] << (8 * i);
            v4_node_port_give(n, 0, (v4_cell)word);
            at++;
        }
    }

    say(b, "PARITY:V4_NUCLEUS name=" NUCLEUS_NAME);
    say(b, " capsule_id="); say_hex(b, cap->capsule_id);
    say(b, " capsule_hash="); say_hex(b, cap->content_hash);
    say(b, " words="); say_dec(b, word_count(n, b->im));
    say(b, "\n");
    return 1;
}

/* ---- the whole boot ------------------------------------------------------ */

int v4_boot_run(const v4_boot *b)
{
    unsigned i;

    post_len = 0; post_seen = 0; post_clean = 0;
    /* a node is born with nothing in it, and is sent the nucleus */
    if (!v4_image_born(b->n, b->es, b->h, b->im)) {
        say(b, "V4: the nucleus is not for this build of the engine\nPARITY:FAIL\nPOST: FAILED\n");
        return 0;
    }
    /* what is on its two ports, the kernel and the console, takes a word
     * when it is written: the node need never wait before writing to them
     * (node.h, v4_node_port_status) */
    v4_node_port_status(b->n, 0, 1u << KERNEL_PORT | 1u << CONSOLE_PORT);
    if (!load_nucleus(b)) { say(b, "PARITY:FAIL\nPOST: FAILED\n"); return 0; }

    /* the kernel's words, before any capsule that might use one */
    for (i = 0; i < b->word_count; i++) {
        char text[64];
        unsigned at = 0, k, num = i + 1, div = 1;
        while (num / div >= 10) div *= 10;
        for (; div; div /= 10) text[at++] = (char)('0' + num / div % 10);
        for (k = 0; " KERNEL-WORD "[k]; k++) text[at++] = " KERNEL-WORD "[k];
        for (k = 0; b->words[i][k] && at < sizeof text; k++) text[at++] = b->words[i][k];
        if (b->words[i][k] || v4_boot_line(b, text, at) != V4_TEXT_COMPLETED) {
            say(b, "V4: kernel word "); say(b, b->words[i]); say(b, " could not be made\nPARITY:FAIL\nPOST: FAILED\n");
            return 0;
        }
    }

    for (i = 0; i < sizeof boot_capsules / sizeof boot_capsules[0]; i++)
        if (!load_capsule(b, boot_capsules[i])) { say(b, "PARITY:FAIL\nPOST: FAILED\n"); return 0; }
#if V4_POST_AT_BOOT
    if (!post_seen || !post_clean) {
        say(b, post_seen ? "V4: POST reported failures\n" : "V4: POST did not report\n");
        say(b, "PARITY:FAIL\nPOST: FAILED\n");
        return 0;
    }
#endif

    /* The kernel's words and the capsules are the system: COLD comes back
     * to here, and FORGET will not go below it. */
    v4_image_seal(b->n, b->im);

#if V4_POST_AT_BOOT
    say(b, "PARITY:OK\nPOST: PASSED\n");
#else
    say(b, "PARITY:OK\n");
#endif
    return 1;
}