write mode: implemented generation of the cpio archive format.

7 files changed, 415 insertions, 18 deletions

diff --git a/include/tpax/tpax.h b/include/tpax/tpax.h
index b811987..1ea9887 100644
--- a/include/tpax/tpax.h
+++ b/include/tpax/tpax.h
@@ -184,6 +184,10 @@ tpax_api int  tpax_output_error_vector      (const struct tpax_driver_ctx *);
 tpax_api int  tpax_output_error_record      (const struct tpax_driver_ctx *, const struct tpax_error_info *);
 
 /* meta interfaces */
+tpax_api int  tpax_meta_init_cpio_header    (const char * pathname, const struct stat *,
+                                             const char * linkname, int c_dev, int c_ino, int c_nlink,
+                                             struct tpax_cpio_header *);
+
 tpax_api int  tpax_meta_init_ustar_header   (const char * pathname, const struct stat *,
                                              const char * linkname, struct tpax_ustar_header *);
 
diff --git a/project/common.mk b/project/common.mk
index 03e35ee..d600ff2 100644
--- a/project/common.mk
+++ b/project/common.mk
@@ -8,6 +8,7 @@ API_SRCS = \
 	src/logic/tpax_archive_reset.c \
 	src/logic/tpax_archive_write.c \
 	src/logic/tpax_queue_vector.c \
+	src/meta/tpax_init_cpio_header.c \
 	src/meta/tpax_init_ustar_header.c \
 	src/output/tpax_output_error.c \
 	src/skin/tpax_skin_default.c \
diff --git a/src/driver/tpax_driver_ctx.c b/src/driver/tpax_driver_ctx.c
index 74db70d..6a5f882 100644
--- a/src/driver/tpax_driver_ctx.c
+++ b/src/driver/tpax_driver_ctx.c
@@ -872,10 +872,6 @@ int tpax_lib_get_driver_ctx(
 			return tpax_driver_error_not_implemented(
 				fdctx->fderr,program,"the pax format",meta);
 
-		case TPAX_DRIVER_WRITE_FORMAT_CPIO:
-			return tpax_driver_error_not_implemented(
-				fdctx->fderr,program,"the cpio format",meta);
-
 		default:
 			break;
 	}
@@ -954,6 +950,9 @@ static void tpax_free_driver_ctx_impl(struct tpax_driver_ctx_alloc * ictx)
 	if (ictx->ctx.direntv)
 		free(ictx->ctx.direntv);
 
+	if (ictx->ctx.cpiov)
+		free(ictx->ctx.cpiov);
+
 	argv_free(ictx->meta);
 	free(ictx);
 }
diff --git a/src/internal/tpax_driver_impl.h b/src/internal/tpax_driver_impl.h
index 7726e1f..a7350d7 100644
--- a/src/internal/tpax_driver_impl.h
+++ b/src/internal/tpax_driver_impl.h
@@ -64,6 +64,9 @@ struct tpax_dirent {
 	int                           fdat;
 	int                           depth;
 	int                           flags;
+	int                           nlink;
+	int                           cpdev;
+	int                           cpino;
 	dev_t                         srdev;
 	dev_t                         stdev;
 	ino_t                         stino;
@@ -107,6 +110,7 @@ struct tpax_driver_ctx_impl {
 	char **                       prefixp;
 	char **                       prefcap;
 	char *                        prefptr[64];
+	struct tpax_dirent **         cpiov;
 	struct tpax_dirent **         direntv;
 	struct tpax_dirent_buffer *   dirents;
 	struct tpax_dirent *          dirmark;
diff --git a/src/logic/tpax_archive_write.c b/src/logic/tpax_archive_write.c
index fbf46a7..492d3b1 100644
--- a/src/logic/tpax_archive_write.c
+++ b/src/logic/tpax_archive_write.c
@@ -13,6 +13,16 @@
 #define ssizeof(x) (ssize_t)(sizeof(x))
 #endif
 
+static const char tpax_cpio_trailer[11] = TPAX_CPIO_TRAILER;
+
+static void tpax_cpio_octal_write(char * ch, ssize_t len, uint64_t val)
+{
+	for (; len; ) {
+		ch[--len] = val % 8 + '0';
+		val /= 8;
+	}
+}
+
 static int tpax_archive_append_memory_data(
 	int                             fdout,
 	const void *                    buf,
@@ -46,6 +56,9 @@ static int tpax_archive_append_pad(
 	ssize_t nbytes;
 	char    buf[512];
 
+	if (dctx->cctx->drvflags & TPAX_DRIVER_WRITE_FORMAT_CPIO)
+		return 0;
+
 	nbytes  = st->st_size;
 	nbytes += 0x1ff;
 	nbytes |= 0x1ff;
@@ -112,6 +125,7 @@ static int tpax_archive_write_impl(
 {
 	int                             ret;
 	struct tpax_unit_ctx *          uctx;
+	struct tpax_cpio_header         chdr;
 	struct tpax_ustar_header        uhdr;
 	const struct stat *             st;
 	struct stat                     stbuf;
@@ -206,9 +220,8 @@ static int tpax_archive_write_impl(
 	tpax_driver_set_ectx(
 		dctx,0,path);
 
-	/* header and data offsets: todo pax and cpio */
+	/* header offset */
 	hpos = tpax_get_driver_cpos(dctx);
-	dpos = hpos + sizeof(uhdr);
 
 	/* header */
 	if (dctx->cctx->drvflags & TPAX_DRIVER_WRITE_FORMAT_RUSTAR) {
@@ -216,6 +229,14 @@ static int tpax_archive_write_impl(
 
 	} else if (dctx->cctx->drvflags & TPAX_DRIVER_WRITE_FORMAT_USTAR) {
 		ret = tpax_meta_init_ustar_header(apath,st,slnk,&uhdr);
+
+	} else if (dctx->cctx->drvflags & TPAX_DRIVER_WRITE_FORMAT_CPIO) {
+		ret = tpax_meta_init_cpio_header(
+			apath,st,slnk,
+			cdent->cpdev,
+			cdent->cpino,
+			cdent->nlink,
+			&chdr);
 	}
 
 	if (ret < 0)
@@ -260,8 +281,28 @@ static int tpax_archive_write_impl(
 		}
 	}
 
-	/* append header */
-	if (tpax_archive_append_memory_data(fdout,&uhdr,ssizeof(uhdr)) < 0) {
+	/* append header (and cpio symbolic link data) */
+	if (dctx->cctx->drvflags & TPAX_DRIVER_WRITE_FORMAT_CPIO) {
+		ret = tpax_archive_append_memory_data(
+			fdout,&chdr,
+			offsetof(struct tpax_cpio_header,c_namedata));
+
+		if (ret == 0)
+			ret = tpax_archive_append_memory_data(
+				fdout,apath,
+				strlen(apath) + 1);
+
+		if ((ret == 0) && slnk)
+			ret = tpax_archive_append_memory_data(
+				fdout,slnk,
+				strlen(slnk));
+	} else {
+		ret = tpax_archive_append_memory_data(
+			fdout,&uhdr,
+			ssizeof(uhdr));
+	}
+
+	if (ret < 0) {
 		if (fdtmp >= 0)
 			close(fdtmp);
 
@@ -270,6 +311,16 @@ static int tpax_archive_write_impl(
 			dctx,uctx);
 	}
 
+	/* data offset */
+	if (dctx->cctx->drvflags & TPAX_DRIVER_WRITE_FORMAT_CPIO) {
+		dpos  = hpos;
+		dpos += offsetof(struct tpax_cpio_header,c_namedata);
+		dpos += strlen(apath) + 1;
+		dpos += slnk ? strlen(slnk) : 0;
+	} else {
+		dpos  = hpos + ssizeof(uhdr);
+	}
+
 	tpax_set_driver_cpos(dctx,dpos);
 
 	/* all done? */
@@ -352,8 +403,79 @@ int tpax_archive_write(const struct tpax_driver_ctx * dctx)
 
 static int tpax_archive_seal_cpio(const struct tpax_driver_ctx * dctx)
 {
-	(void)dctx;
-	return -1;
+	int                             fdout;
+	off_t                           cpos;
+	ssize_t                         nbytes;
+	ssize_t                         blksize;
+	ssize_t                         nwritten;
+	char                            buf[1024];
+	struct tpax_cpio_header         chdr;
+
+	/* trailer initialization */
+	memset(buf,0,sizeof(buf));
+	memset(&chdr,0,sizeof(chdr));
+
+	tpax_cpio_octal_write(chdr.c_magic,sizeof(chdr.c_magic),0070707);
+	tpax_cpio_octal_write(chdr.c_dev,sizeof(chdr.c_dev),0);
+	tpax_cpio_octal_write(chdr.c_ino,sizeof(chdr.c_ino),0);
+	tpax_cpio_octal_write(chdr.c_mode,sizeof(chdr.c_mode),0);
+	tpax_cpio_octal_write(chdr.c_uid,sizeof(chdr.c_uid),0);
+	tpax_cpio_octal_write(chdr.c_gid,sizeof(chdr.c_gid),0);
+	tpax_cpio_octal_write(chdr.c_nlink,sizeof(chdr.c_nlink),1);
+	tpax_cpio_octal_write(chdr.c_rdev,sizeof(chdr.c_rdev),0);
+	tpax_cpio_octal_write(chdr.c_mtime,sizeof(chdr.c_mtime),0);
+
+	tpax_cpio_octal_write(chdr.c_filesize,sizeof(chdr.c_filesize),0);
+	tpax_cpio_octal_write(chdr.c_namesize,sizeof(chdr.c_namesize),sizeof(tpax_cpio_trailer));
+
+	/* fdout, archive block size, current position */
+	fdout   = tpax_driver_fdout(dctx);
+	blksize = dctx->cctx->blksize;
+	cpos    = tpax_get_driver_cpos(dctx);
+
+	/* trailer header */
+	nbytes = offsetof(struct tpax_cpio_header,c_namedata);
+
+	if (tpax_archive_append_memory_data(fdout,&chdr,nbytes) < 0)
+		return TPAX_SYSTEM_ERROR(dctx);
+
+	cpos += nbytes;
+
+	/* trailer c_namedata[] */
+	nbytes = sizeof(tpax_cpio_trailer);
+
+	if (tpax_archive_append_memory_data(fdout,tpax_cpio_trailer,nbytes) < 0)
+		return TPAX_SYSTEM_ERROR(dctx);
+
+	cpos += nbytes;
+
+	/* pad the current block to a 512 byte boundary */
+	nbytes = (cpos % 512) ? 512 - (cpos % 512) : 0;
+
+	if (tpax_archive_append_memory_data(fdout,buf,nbytes) < 0)
+		return TPAX_SYSTEM_ERROR(dctx);
+
+	cpos += nbytes;
+
+	/* already at block boundary? */
+	if ((cpos % blksize) == 0) {
+		tpax_set_driver_cpos(dctx,cpos);
+		return 0;
+	}
+
+	/* zero-fill the remainder of the block */
+	nbytes  = cpos / blksize;
+	nbytes *= blksize;
+	nbytes += blksize;
+
+	for (nwritten=cpos; nwritten<nbytes; nwritten+=512)
+		if (tpax_archive_append_memory_data(fdout,buf,512) < 0)
+			return TPAX_SYSTEM_ERROR(dctx);
+
+	/* all done */
+	tpax_set_driver_cpos(dctx,nwritten);
+
+	return 0;
 }
 
 int tpax_archive_seal(const struct tpax_driver_ctx * dctx)
@@ -365,6 +487,10 @@ int tpax_archive_seal(const struct tpax_driver_ctx * dctx)
 	ssize_t blksize;
 	char    buf[1024];
 
+	/* cpio trailer? */
+	if (dctx->cctx->drvflags & TPAX_DRIVER_WRITE_FORMAT_CPIO)
+		return tpax_archive_seal_cpio(dctx);
+
 	/* archive block size, current position */
 	blksize = dctx->cctx->blksize;
 	cpos    = tpax_get_driver_cpos(dctx);
@@ -373,10 +499,6 @@ int tpax_archive_seal(const struct tpax_driver_ctx * dctx)
 	if (cpos % 512)
 		return TPAX_CUSTOM_ERROR(dctx,TPAX_ERR_FLOW_ERROR);
 
-	/* cpio trailer? */
-	if (dctx->cctx->drvflags & TPAX_DRIVER_WRITE_FORMAT_CPIO)
-		return tpax_archive_seal_cpio(dctx);
-
 	/* ustar, pax */
 	fdout = tpax_driver_fdout(dctx);
 	memset(buf,0,sizeof(buf));
diff --git a/src/logic/tpax_queue_vector.c b/src/logic/tpax_queue_vector.c
index cae4da0..1b10feb 100644
--- a/src/logic/tpax_queue_vector.c
+++ b/src/logic/tpax_queue_vector.c
@@ -53,6 +53,118 @@ tpax_hidden const char * tpax_queue_item_full_path(
 	return pathbuf;
 }
 
+static int tpax_cpio_dirent_compare(const void * a, const void * b)
+{
+	const struct tpax_dirent * direnta;
+	const struct tpax_dirent * direntb;
+
+	direnta = *(const struct tpax_dirent **)a;
+	direntb = *(const struct tpax_dirent **)b;
+
+	return (direnta->stdev == direntb->stdev)
+		? direnta->stino - direntb->stino
+		: direnta->stdev - direntb->stdev;
+}
+
+static int tpax_update_cpio_queue_vector(const struct tpax_driver_ctx * dctx)
+{
+	struct tpax_driver_ctx_impl *   ictx;
+	struct tpax_dirent **           cpiov;
+	struct tpax_dirent **           direntv;
+	struct tpax_dirent *            cdent;
+	dev_t                           stdev;
+	ino_t                           stino;
+	int                             cpdev;
+	int                             cpino;
+	int                             nlink;
+	int                             idx;
+
+	/* driver */
+	ictx = tpax_get_driver_ictx(dctx);
+
+	/* not needed? */
+	if (ictx->nqueued == 0)
+		return 0;
+
+	/* vector copy */
+	direntv = ictx->direntv;
+	cpiov   = ictx->cpiov;
+
+	for (; *direntv; )
+		*cpiov++ = *direntv++;
+
+	/* sort by real st_dev, st_ino */
+	qsort(ictx->cpiov,ictx->nqueued,sizeof(*cpiov),tpax_cpio_dirent_compare);
+
+	/* set the cpio internal c_dev, c_ino, and c_nlink fields (U+1F620) */
+	cpiov = ictx->cpiov;
+	cdent = cpiov[0];
+	cpiov++;
+
+	cpdev = 1;
+	cpino = 1;
+	nlink = 1;
+
+	stdev = cdent->stdev;
+	stino = cdent->stino;
+
+	cdent->cpdev = cpdev;
+	cdent->cpino = cpino;
+	cdent->nlink = nlink;
+
+	for (; *cpiov; ) {
+		cdent = *cpiov;
+
+		if (cdent->srdev > 0777777)
+			return TPAX_CUSTOM_ERROR(
+				dctx,
+				TPAX_ERR_FLOW_ERROR);
+
+		if ((cdent->stdev == stdev) && (cdent->stino == stino)) {
+			nlink++;
+
+		} else if (cdent->stdev > stdev) {
+			if (nlink > 1)
+				for (idx=2; idx<=nlink; idx++)
+					cpiov[-idx]->nlink = nlink;
+
+			cpdev++;
+			cpino = 1;
+			nlink = 1;
+
+			stdev = cdent->stdev;
+			stino = cdent->stino;
+
+			if (cpdev > 0777777)
+				return TPAX_CUSTOM_ERROR(
+					dctx,
+					TPAX_ERR_FLOW_ERROR);
+
+		} else if (cdent->stino > stino) {
+			if (nlink > 1)
+				for (idx=2; idx<=nlink; idx++)
+					cpiov[-idx]->nlink = nlink;
+
+			cpino++;
+			stino = cdent->stino;
+			nlink = 1;
+
+			if (cpino > 0777777)
+				return TPAX_CUSTOM_ERROR(
+					dctx,
+					TPAX_ERR_FLOW_ERROR);
+		}
+
+		cdent->cpdev = cpdev;
+		cdent->cpino = cpino;
+		cdent->nlink = nlink;
+
+		cpiov++;
+	}
+
+	return 0;
+}
+
 tpax_hidden int tpax_update_queue_vector(const struct tpax_driver_ctx * dctx)
 {
 	uintptr_t                       addr;
@@ -67,20 +179,26 @@ tpax_hidden int tpax_update_queue_vector(const struct tpax_driver_ctx * dctx)
 	ictx = tpax_get_driver_ictx(dctx);
 
 	/* vector alloc */
-	if (ictx->direntv)
+	if (ictx->direntv) {
 		free(ictx->direntv);
+		free(ictx->cpiov);
+
+		ictx->direntv = 0;
+		ictx->cpiov   = 0;
+	}
 
 	arrsize = ictx->nqueued + 1;
 
 	if (!(ictx->direntv = calloc(arrsize,sizeof(struct tpax_dirent *))))
 		return TPAX_SYSTEM_ERROR(dctx);
 
-	if (ictx->nqueued == 0)
-		return 0;
+	if (dctx->cctx->drvflags & TPAX_DRIVER_WRITE_FORMAT_CPIO)
+		if (!(ictx->cpiov = calloc(arrsize,sizeof(struct tpax_dirent *))))
+			return TPAX_SYSTEM_ERROR(dctx);
 
 	/* queue vector */
 	dentbuf = tpax_get_driver_dirents(dctx);
-	cdent   = dentbuf->dbuf;
+	cdent   = ictx->nqueued ? dentbuf->dbuf : 0;
 
 	for (direntv=ictx->direntv; cdent; direntv++) {
 		*direntv = cdent;
@@ -97,5 +215,9 @@ tpax_hidden int tpax_update_queue_vector(const struct tpax_driver_ctx * dctx)
 		cdent = cnext;
 	}
 
+	if (dctx->cctx->drvflags & TPAX_DRIVER_WRITE_FORMAT_CPIO)
+		if (tpax_update_cpio_queue_vector(dctx) < 0)
+			return TPAX_NESTED_ERROR(dctx);
+
 	return 0;
 }
diff --git a/src/meta/tpax_init_cpio_header.c b/src/meta/tpax_init_cpio_header.c
new file mode 100644
index 0000000..3de22f6
--- /dev/null
+++ b/src/meta/tpax_init_cpio_header.c
@@ -0,0 +1,145 @@
+/**************************************************************/
+/*  tpax: a topological pax implementation                    */
+/*  Copyright (C) 2020--2024  SysDeer Technologies, LLC       */
+/*  Released under GPLv2 and GPLv3; see COPYING.TPAX.         */
+/**************************************************************/
+
+#include <stdint.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <errno.h>
+#include <grp.h>
+#include <pwd.h>
+#include <sys/stat.h>
+
+#include <tpax/tpax.h>
+#include <tpax/tpax_specs.h>
+#include "tpax_driver_impl.h"
+
+#ifndef ssizeof
+#define ssizeof(x) (ssize_t)(sizeof(x))
+#endif
+
+#define TPAX_CPIO_PERM_MASK         \
+	( S_ISUID | S_ISGID          \
+	| S_IRUSR | S_IWUSR | S_IXUSR \
+	| S_IRGRP | S_IWGRP | S_IXGRP \
+	| S_IROTH | S_IWOTH | S_IXOTH )
+
+static void tpax_octal_write(char * ch, ssize_t len, uint64_t val)
+{
+	for (; len; ) {
+		ch[--len] = val % 8 + '0';
+		val /= 8;
+	}
+}
+
+int tpax_meta_init_cpio_header(
+	const char *                    path,
+	const struct stat *             st,
+	const char *                    linkname,
+	int                             cdev,
+	int                             cino,
+	int                             cnlink,
+	struct tpax_cpio_header *       chdr)
+{
+	size_t		fnsize;
+	size_t		lnklen;
+	size_t		stsize;
+	int64_t		stmtim;
+	uint32_t	typeflag;
+	uint32_t	permbits;
+	uint32_t	modebits;
+
+	/* filename size */
+	fnsize = strlen(path) + 1;
+
+	/* size & mtime validation */
+	stsize = S_ISREG(st->st_mode) ? st->st_size : 0;
+	stmtim = st->st_mtim.tv_sec;
+
+	if (stsize > 077777777777)
+		return -1;
+
+	if ((stmtim < 0) || (stmtim > 077777777777))
+		return -1;
+
+	/* linkname validation */
+	if (S_ISLNK(st->st_mode) && !linkname)
+		return -1;
+
+	lnklen = S_ISLNK(st->st_mode)
+		? strlen(linkname)
+		: 0;
+
+	/* typeflag validation */
+	if (S_ISREG(st->st_mode))
+		typeflag = TPAX_CPIO_FILEMODE_ISREG;
+	else if (S_ISLNK(st->st_mode))
+		typeflag = TPAX_CPIO_FILEMODE_ISLNK;
+	else if (S_ISDIR(st->st_mode))
+		typeflag = TPAX_CPIO_FILEMODE_ISDIR;
+	else if (S_ISCHR(st->st_mode))
+		typeflag = TPAX_CPIO_FILEMODE_ISCHR;
+	else if (S_ISBLK(st->st_mode))
+		typeflag = TPAX_CPIO_FILEMODE_ISBLK;
+	else if (S_ISFIFO(st->st_mode))
+		typeflag = TPAX_CPIO_FILEMODE_ISFIFO;
+	else if (S_ISSOCK(st->st_mode))
+		typeflag = TPAX_CPIO_FILEMODE_ISSOCK;
+	else
+		return -1;
+
+	/* permbits, modeflag */
+	permbits = st->st_mode & TPAX_CPIO_PERM_MASK;
+	modebits = permbits | typeflag;
+
+	/* one shot */
+	memset(chdr,0,sizeof(*chdr));
+
+	/* c_magic */
+	chdr->c_magic[0] = '0';
+	chdr->c_magic[1] = '7';
+	chdr->c_magic[2] = '0';
+	chdr->c_magic[3] = '7';
+	chdr->c_magic[4] = '0';
+	chdr->c_magic[5] = '7';
+
+	/* c_dev, c_ino */
+	tpax_octal_write(chdr->c_dev,ssizeof(chdr->c_dev),cdev);
+	tpax_octal_write(chdr->c_ino,ssizeof(chdr->c_ino),cino);
+
+	/* c_mode */
+	tpax_octal_write(chdr->c_mode,ssizeof(chdr->c_mode),modebits);
+
+	/* c_uid, c_gid */
+	if ((uint64_t)st->st_uid <= 0777777)
+		tpax_octal_write(chdr->c_uid,ssizeof(chdr->c_uid),st->st_uid);
+	else
+		tpax_octal_write(chdr->c_uid,ssizeof(chdr->c_uid),0);
+
+	if ((uint64_t)st->st_gid <= 0777777)
+		tpax_octal_write(chdr->c_gid,ssizeof(chdr->c_gid),st->st_gid);
+	else
+		tpax_octal_write(chdr->c_gid,ssizeof(chdr->c_gid),0);
+
+	/* c_nlink */
+	tpax_octal_write(chdr->c_nlink,ssizeof(chdr->c_nlink),cnlink);
+
+	/* c_rdev */
+	tpax_octal_write(chdr->c_rdev,ssizeof(chdr->c_rdev),st->st_rdev);
+
+	/* c_mtime */
+	tpax_octal_write(chdr->c_mtime,ssizeof(chdr->c_mtime),stmtim);
+
+	/* c_namesize */
+	tpax_octal_write(chdr->c_namesize,ssizeof(chdr->c_namesize),fnsize);
+
+	/* c_filesize */
+	tpax_octal_write(chdr->c_filesize,ssizeof(chdr->c_filesize),lnklen ? lnklen : stsize);
+
+	/* all done; c_name to be written along with c_filedata */
+	return 0;
+}