What causes "gencmn: udata_finish() failed - %s\n" in nodejs/node?

Free space on the output volume and re-run (this is usually the cause of a finish/close failure).

gencmn: udata_finish() failed - %s\n

Error message

gencmn: udata_finish() failed - %s\n

What it means

After writing all data and padding, udata_finish(out, &errorCode) reported U_FAILURE. gencmn prints udata_finish() failed with u_errorName and exit(errorCode), indicating the output container could not be finalized/flushed.

Source

Thrown at deps/icu-small/source/tools/toolutil/pkg_gencmn.cpp:306

            T_FileStream_close(file);
            length=files[i].fileSize;

            if (nread != files[i].fileSize) {
              fprintf(stderr, "gencmn: unable to read %s properly (got %ld/%ld byte%s)\n", files[i].pathname,  (long)nread, (long)files[i].fileSize, files[i].fileSize == 1 ? "" : "s");
                exit(U_FILE_ACCESS_ERROR);
            }
        }

        /* pad to 16-align the last file (cleaner, avoids growing .dat files in icuswap) */
        length&=0xf;
        if(length!=0) {
            udata_writePadding(out, 16-length);
        }

        /* finish */
        udata_finish(out, &errorCode);
        if(U_FAILURE(errorCode)) {
            fprintf(stderr, "gencmn: udata_finish() failed - %s\n", u_errorName(errorCode));
            exit(errorCode);
        }
    } else {
        /* write a .c source file with the table of contents */
        char *filename;
        FileStream *out;

        /* create the output filename */
        filename=s=buffer;
        uprv_strcpy(filename, destDir);
        s=filename+uprv_strlen(filename);
        if(s>filename && *(s-1)!=U_FILE_SEP_CHAR) {
            *s++=U_FILE_SEP_CHAR;
        }
        uprv_strcpy(s, name);
        if(*(type)!=0) {
            s+=uprv_strlen(s);
            *s++='_';

View on GitHub (pinned to 1b2de5e052)

Solutions

Free space on the output volume and re-run (this is usually the cause of a finish/close failure).
Run the build on a local disk instead of a network mount.
Remove the partial output .dat so gencmn starts fresh: `rm -f "$DEST/$NAME.$TYPE"`.
Check dmesg/system logs for I/O errors on the target device.

Example fix

# before: output on a near-full network mount
gencmn -d /mnt/nfs/out ...
# after
df -h ./out && gencmn -d ./out ...   # local, enough free space

Defensive patterns

Strategy: validation

Validate before calling

# check free space on the destination volume before running
df -k "$DEST" | awk 'NR==2 && $4 < 1048576 { exit 1 }' || { echo "low disk on $DEST" >&2; exit 1; }

Prevention

Ensure adequate free space (>= 2x expected .dat size) before building.
Build on local disk; network mounts can drop at close.
Check gencmn exit code and remove any partial .dat on failure.
Watch dmesg for I/O errors if this recurs.

When it happens

Trigger: A write/flush error while finalizing the .dat header or closing the file: disk full at the very end, I/O error on the output device, or an internal udata accounting failure. The check is `if(U_FAILURE(errorCode))` immediately after udata_finish.

Common situations: Disk fills up during the write so the final flush/header rewrite fails; output on a network volume that drops at close; an interrupted build leaving a partial .dat.

Related errors

AI-assisted analysis of nodejs/node@1b2de5e052 (2026-08-13). Data as JSON: /api/errors/7242baa2f385adde. Report an issue: GitHub.