path: root/checksum.c

/*
 * checksum.c - SHA256 source verification and update
 * https://rawnix.org
 *
 * Copyright (c) 2025 zorz@gmx.com
 * SPDX-License-Identifier: BSD-2-Clause
 */

#include "mkpkg.h"

#include <err.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <strings.h>
#include <sys/stat.h>
#include <unistd.h>

#include <openssl/evp.h>


/* ── SHA256 ───────────────────────────────────────────────────────────── */

/*
 * Compute hex SHA256 of a file into result[65].
 * Returns 0 on success, -1 on failure.
 *
 * 1 MB read buffer: with the old 8 KB buffer a 4 GB rocm-class
 * package meant ~500k fread calls; large sequential reads let the
 * kernel readahead do its job.  Heap-allocated — 1 MB does not
 * belong on the stack.
 */
#define SHA_BUF_SIZE (1024 * 1024)

static int sha256_file(const char *path, char *result, size_t result_size)
{
    FILE        *fp;
    EVP_MD_CTX  *ctx;
    unsigned char *buf;
    unsigned char digest[EVP_MAX_MD_SIZE];
    unsigned int  digest_len;
    size_t        n;
    unsigned int  i;

    if (result_size < 65) return -1;

    fp = fopen(path, "rb");
    if (!fp) return -1;

    buf = malloc(SHA_BUF_SIZE);
    if (!buf) { fclose(fp); return -1; }

    ctx = EVP_MD_CTX_new();
    if (!ctx) { free(buf); fclose(fp); return -1; }

    if (EVP_DigestInit_ex(ctx, EVP_sha256(), NULL) != 1) goto fail;

    while ((n = fread(buf, 1, SHA_BUF_SIZE, fp)) > 0) {
        if (EVP_DigestUpdate(ctx, buf, n) != 1) goto fail;
    }

    if (ferror(fp)) goto fail;
    if (EVP_DigestFinal_ex(ctx, digest, &digest_len) != 1) goto fail;

    EVP_MD_CTX_free(ctx);
    free(buf);
    fclose(fp);

    for (i = 0; i < digest_len; i++)
        snprintf(result + i * 2, result_size - i * 2, "%02x", digest[i]);
    result[digest_len * 2] = '\0';
    return 0;

fail:
    EVP_MD_CTX_free(ctx);
    free(buf);
    fclose(fp);
    return -1;
}

/* ── Streaming SHA256 + build-time hash cache ─────────────────────────── */

/*
 * Used by mkpkg to hash the package WHILE the compressor writes it
 * (see arc_create_pkg's sink), instead of re-reading a multi-GB
 * tarball from disk just to sign data that flowed through the
 * machine seconds earlier.  After a big rocm-class build the page
 * cache is usually evicted, so that re-read was frequently a real
 * full disk pass.
 */

void *sha256_stream_new(void)
{
    EVP_MD_CTX *ctx = EVP_MD_CTX_new();

    if (!ctx)
        return NULL;
    if (EVP_DigestInit_ex(ctx, EVP_sha256(), NULL) != 1) {
        EVP_MD_CTX_free(ctx);
        return NULL;
    }
    return ctx;
}

void sha256_stream_update(void *vctx, const void *buf, size_t len)
{
    if (vctx)
        EVP_DigestUpdate((EVP_MD_CTX *)vctx, buf, len);
}

/* Finalise into result[65] (hex) and free the context. */
int sha256_stream_final(void *vctx, char *result, size_t result_size)
{
    EVP_MD_CTX   *ctx = vctx;
    unsigned char digest[EVP_MAX_MD_SIZE];
    unsigned int  digest_len;
    unsigned int  i;

    if (!ctx || result_size < 65) {
        if (ctx)
            EVP_MD_CTX_free(ctx);
        return -1;
    }

    if (EVP_DigestFinal_ex(ctx, digest, &digest_len) != 1) {
        EVP_MD_CTX_free(ctx);
        return -1;
    }
    EVP_MD_CTX_free(ctx);

    for (i = 0; i < digest_len; i++)
        snprintf(result + i * 2, result_size - i * 2, "%02x", digest[i]);
    result[digest_len * 2] = '\0';
    return 0;
}

/*
 * One-slot cache: the hash of the package mkpkg just wrote.
 * make_sig_checksums() consults it instead of re-hashing.  The stat
 * snapshot (size + mtime) guards against the file changing between
 * build and sign — if it doesn't match exactly, fall back to a
 * fresh sha256_file() read.  Single-threaded by design, like the
 * rest of mkpkg.
 */
static struct {
    char   path[PATH_MAX_LEN];
    char   sha[65];
    off_t  size;
    time_t mtime;
    int    valid;
} g_sha_cache;

void sha256_cache_store(const char *path, const char *sha)
{
    struct stat st;

    g_sha_cache.valid = 0;

    if (!path || !sha || stat(path, &st) != 0)
        return;

    snprintf(g_sha_cache.path, sizeof(g_sha_cache.path), "%s", path);
    snprintf(g_sha_cache.sha,  sizeof(g_sha_cache.sha),  "%s", sha);
    g_sha_cache.size  = st.st_size;
    g_sha_cache.mtime = st.st_mtime;
    g_sha_cache.valid = 1;
}

static int sha256_cached(const char *path, char *result, size_t result_size)
{
    struct stat st;

    if (!g_sha_cache.valid || result_size < 65)
        return -1;
    if (strcmp(path, g_sha_cache.path) != 0)
        return -1;
    if (stat(path, &st) != 0 ||
        st.st_size  != g_sha_cache.size ||
        st.st_mtime != g_sha_cache.mtime)
        return -1;

    snprintf(result, result_size, "%s", g_sha_cache.sha);
    return 0;
}

/* ── Local filename resolution ────────────────────────────────────────── */

static char *local_path(const source_t *src, const config_t *config)
{
    char path[PATH_MAX_LEN];
    const char *url = src->url;

    if (src->rename && strcmp(src->rename, "SKIP") != 0) {
        snprintf(path, sizeof(path), "%s/%s",
                 config->source_dir, src->rename);
    } else if (strchr(url, '/')) {
        const char *base = strrchr(url, '/');
        snprintf(path, sizeof(path), "%s/%s",
                 config->source_dir, base ? base + 1 : url);
    } else {
        snprintf(path, sizeof(path), "%s/%s",
                 config->root_dir, url);
    }

    return xstrdup(path);
}

static const char *path_basename(const char *p)
{
    const char *s = strrchr(p, '/');
    return s ? s + 1 : p;
}

/* ── verify_checksums ─────────────────────────────────────────────────── */

int verify_checksums(pkgfile_t *pkgfile, config_t *config)
{
    int errors = 0;
    int i;

    if (pkgfile->sha256sum_count == 0)
        return 0;

    for (i = 0; i < pkgfile->source_count; i++) {
        source_t   *src  = &pkgfile->sources[i];
        char       *fp   = local_path(src, config);
        const char *base = path_basename(fp);
        char        computed[65];

        if (i >= pkgfile->sha256sum_count) { free(fp); continue; }
        if (strcmp(pkgfile->sha256sums[i], "SKIP") == 0) { free(fp); continue; }

        if (sha256_file(fp, computed, sizeof(computed)) != 0) {
            fprintf(stderr, "mkpkg: error: failed to compute SHA256: %s\n", base);
            errors++;
            free(fp);
            continue;
        }

        if (strcasecmp(computed, pkgfile->sha256sums[i]) != 0) {
            fprintf(stderr, "mkpkg: error: SHA256 mismatch: %s\n", base);
            warnx("  expected: %s", pkgfile->sha256sums[i]);
            warnx("  computed: %s", computed);
            errors++;
        } else {
            info("  %s (SHA256) ok", base);
        }

        free(fp);
    }

    return errors > 0 ? E_CHECKSUM : 0;
}

/* ── generate_checksums ───────────────────────────────────────────────── */

int generate_checksums(pkgfile_t *pkgfile, config_t *config)
{
    int i;

    printf("\nsha256sums=(\n");

    for (i = 0; i < pkgfile->source_count; i++) {
        char *fp = local_path(&pkgfile->sources[i], config);
        char  computed[65];

        if (access(fp, F_OK) != 0 ||
            sha256_file(fp, computed, sizeof(computed)) != 0)
            printf("    \"SKIP\"\n");
        else
            printf("    \"%s\"\n", computed);

        free(fp);
    }

    printf(")\n");
    return 0;
}

/* ── update_checksums ─────────────────────────────────────────────────── */

/*
 * Build the new sha256sums=(...) block into a dynamic buffer.
 * Returns heap-allocated string — caller must free().
 */
static char *build_checksums_block(pkgfile_t *pkgfile, config_t *config)
{
    size_t  cap = 4096;
    size_t  len = 0;
    char   *buf = xmalloc(cap);
    int     i;

#define BPRINTF(fmt, ...) \
    do { \
        int _n; \
        while (1) { \
            _n = snprintf(buf + len, cap - len, fmt, ##__VA_ARGS__); \
            if (_n < 0) { free(buf); return NULL; } \
            if ((size_t)_n < cap - len) { len += (size_t)_n; break; } \
            cap *= 2; \
            buf = xrealloc(buf, cap); \
        } \
    } while (0)

    BPRINTF("sha256sums=(\n");

    for (i = 0; i < pkgfile->source_count; i++) {
        char *fp = local_path(&pkgfile->sources[i], config);
        char  computed[65];

        if (access(fp, F_OK) != 0 ||
            sha256_file(fp, computed, sizeof(computed)) != 0)
            BPRINTF("    \"SKIP\"\n");
        else
            BPRINTF("    \"%s\"\n", computed);

        free(fp);
    }

    BPRINTF(")\n");

#undef BPRINTF

    return buf;
}

/*
 * Read an entire file into a heap buffer.
 * Returns pointer + sets *size. Caller must free().
 */
static char *read_file(const char *path, size_t *size)
{
    FILE   *fp;
    char   *buf;
    size_t  cap = 16384, len = 0;
    size_t  n;

    fp = fopen(path, "r");
    if (!fp) return NULL;

    buf = xmalloc(cap);

    while ((n = fread(buf + len, 1, cap - len - 1, fp)) > 0) {
        len += n;
        if (len + 1 >= cap) {
            cap *= 2;
            buf = xrealloc(buf, cap);
        }
    }

    buf[len] = '\0';
    fclose(fp);
    *size = len;
    return buf;
}

/*
 * Write buf to path atomically via a tmp file in the same directory.
 */
static int write_file_atomic(const char *path, const char *buf, size_t len)
{
    char  tmp[PATH_MAX_LEN];
    FILE *fp;

    snprintf(tmp, sizeof(tmp), "%s.tmp", path);
    fp = fopen(tmp, "w");
    if (!fp) return -1;

    if (fwrite(buf, 1, len, fp) != len || fflush(fp) != 0) {
        fclose(fp);
        unlink(tmp);
        return -1;
    }
    fsync(fileno(fp));   /* make data durable before the rename */
    fclose(fp);

    if (rename(tmp, path) != 0) {
        unlink(tmp);
        return -1;
    }
    return 0;
}

int update_checksums(pkgfile_t *pkgfile, config_t *config)
{
    char   pkgfile_path[PATH_MAX_LEN];
    char  *content, *newblock, *result;
    size_t content_len, newblock_len, result_len;
    char  *start, *end;

    snprintf(pkgfile_path, sizeof(pkgfile_path),
             "%s/%s", config->root_dir, config->pkgfile);

    content = read_file(pkgfile_path, &content_len);
    if (!content)
        err(E_GENERAL, "cannot read: %s", pkgfile_path);

    newblock = build_checksums_block(pkgfile, config);
    if (!newblock) {
        free(content);
        return E_GENERAL;
    }
    newblock_len = strlen(newblock);

    start = strstr(content, "sha256sums=(");

    if (!start) {
        /* insert before build() or append at end */
        char *build_pos = strstr(content, "build()");
        size_t before_len = build_pos ? (size_t)(build_pos - content)
                                      : content_len;
        size_t after_len  = content_len - before_len;

        result_len = before_len + newblock_len + 1 + after_len;
        result     = xmalloc(result_len + 1);

        memcpy(result, content, before_len);
        memcpy(result + before_len, newblock, newblock_len);
        result[before_len + newblock_len] = '\n';
        memcpy(result + before_len + newblock_len + 1,
               content + before_len, after_len);
        result[result_len] = '\0';
    } else {
        /* find the closing ) of sha256sums=(...) */
        int paren = 0;
        int entered = 0;
        end = start;
        while (*end) {
            if (*end == '(') { paren++; entered = 1; }
            if (*end == ')') { paren--; }
            end++;
            if (entered && paren == 0) break;
        }
        /* skip trailing newline */
        if (*end == '\n') end++;

        size_t before_len = (size_t)(start - content);
        size_t after_len  = (size_t)(content + content_len - end);

        result_len = before_len + newblock_len + after_len;
        result     = xmalloc(result_len + 1);

        memcpy(result, content, before_len);
        memcpy(result + before_len, newblock, newblock_len);
        memcpy(result + before_len + newblock_len, end, after_len);
        result[result_len] = '\0';
    }

    free(content);
    free(newblock);

    if (write_file_atomic(pkgfile_path, result, result_len) != 0) {
        free(result);
        err(E_GENERAL, "cannot write: %s", pkgfile_path);
    }

    free(result);
    info("sha256sums updated in %s", config->pkgfile);
    return 0;
}

/* ── make_sig_checksums ───────────────────────────────────────────────── */

int make_sig_checksums(pkgfile_t *pkgfile, config_t *config,
                       const char *outfile)
{
    FILE  *fp;
    char   path[PATH_MAX_LEN];
    char   sha[65];
    struct stat st;
    int    i;

    fp = fopen(outfile, "w");
    if (!fp)
        err(E_SIGNATURE, "cannot create: %s", outfile);

    /*
     * The recipe and footprint lines name the ROLE, not the file:
     * always "MAKEPKG" and ".footprint", even when a locked port hashes
     * MAKEPKG.lock and .footprint.lock.  A lock snapshot of upstream's
     * .signature therefore still verifies the unedited lock recipe, and
     * bind_sig_checksums() maps the names back through config.
     */

    /* MAKEPKG */
    snprintf(path, sizeof(path), "%s/%s",
             config->root_dir, config->pkgfile);
    if (sha256_file(path, sha, sizeof(sha)) == 0)
        fprintf(fp, "SHA256 (MAKEPKG) = %s\n", sha);

    /* .footprint if present */
    snprintf(path, sizeof(path), "%s/%s",
             config->root_dir, config->footprint_file);
    if (stat(path, &st) == 0 && sha256_file(path, sha, sizeof(sha)) == 0)
        fprintf(fp, "SHA256 (.footprint) = %s\n", sha);

    /* source files */
    for (i = 0; i < pkgfile->source_count; i++) {
        source_t *src = &pkgfile->sources[i];
        char     *lp;

        if (src->rename && strcmp(src->rename, "SKIP") == 0) continue;

        lp = local_path(src, config);
        if (sha256_file(lp, sha, sizeof(sha)) == 0)
            fprintf(fp, "SHA256 (%s) = %s\n", path_basename(lp), sha);
        free(lp);
    }

    /*
     * Binary package, if a built tarball exists alongside MAKEPKG.
     *
     * Filename is fully deterministic from MAKEPKG + the configured
     * compression mode — no glob, so a stale tarball with a different
     * version on disk doesn't pollute the manifest.  The line is
     * consumed by the bulk tool on the client side to verify the
     * downloaded binary against the (signify-signed) sum recorded
     * here.  If the binary doesn't exist (e.g. running -us right
     * after -x without a rebuild), the line is just omitted; the
     * existing -us behaviour where it always builds first means
     * this is rarely the case.
     */
    {
        const char *ext = config->compression[0] ? config->compression : "gz";
        char binfile[PATH_MAX_LEN];
        snprintf(binfile, sizeof(binfile), "%s#%s-%s.pkg.tar.%s",
                 pkgfile->name, pkgfile->version, pkgfile->release, ext);
        snprintf(path, sizeof(path), "%s/%s", config->package_dir, binfile);
        /* prefer the hash computed while the package was written;
         * re-read only if the cache is missing or the file changed */
        if (stat(path, &st) == 0 &&
            (sha256_cached(path, sha, sizeof(sha)) == 0 ||
             sha256_file(path, sha, sizeof(sha)) == 0))
            fprintf(fp, "SHA256 (%s) = %s\n", binfile, sha);
    }

    fclose(fp);
    return 0;
}

/* ── bind_sig_checksums ───────────────────────────────────────────────── */

/*
 * The inverse of make_sig_checksums(), run on the checksum list AFTER
 * signify has authenticated it.  `signify -V` proves the list is
 * signed; it says nothing about the files on disk.  Without this, a
 * MAKEPKG with a modified build(), sources and sha256sums=() passed
 * verification against the maintainer's untouched .signature.
 *
 *   MAKEPKG     exactly one line; must match config->pkgfile
 *               (MAKEPKG.lock for a locked port)
 *   .footprint  if listed, config->footprint_file must exist and match
 *
 * Sources need no check here: MAKEPKG is bound, so its sha256sums=()
 * bind them, and verify_checksums() checks those after download —
 * verification runs before download, so the files may not exist yet.
 * The binary line is bulk's: it verifies the fetched package against it.
 */
int bind_sig_checksums(const char *cks_path, config_t *config)
{
    FILE *fp;
    char  line[LINE_MAX_LEN];
    int   recipe_lines = 0, errors = 0;

    fp = fopen(cks_path, "r");
    if (!fp) {
        warn("%s", cks_path);
        return E_SIGNATURE;
    }

    while (fgets(line, sizeof(line), fp)) {
        char        name[256], want[65], got[65], path[PATH_MAX_LEN];
        const char *file;

        if (sscanf(line, "SHA256 (%255[^)]) = %64s", name, want) != 2)
            continue;

        if (strcmp(name, "MAKEPKG") == 0) {
            file = config->pkgfile;
            recipe_lines++;
        } else if (strcmp(name, ".footprint") == 0) {
            file = config->footprint_file;
        } else {
            continue;
        }

        snprintf(path, sizeof(path), "%s/%s", config->root_dir, file);
        if (sha256_file(path, got, sizeof(got)) != 0) {
            fprintf(stderr, "mkpkg: error: %s is covered by the "
                    "signature but cannot be read\n", file);
            errors++;
        } else if (strcasecmp(got, want) != 0) {
            fprintf(stderr, "mkpkg: error: %s does not match its "
                    "signed checksum\n", file);
            errors++;
        }
    }
    fclose(fp);

    if (recipe_lines != 1) {
        fprintf(stderr, "mkpkg: error: signature %s the recipe "
                "(expected one MAKEPKG line)\n",
                recipe_lines ? "lists more than one" : "does not cover");
        errors++;
    }

    return errors ? E_SIGNATURE : 0;
}