lib9p: Split internal.h up, rename some CONFIG symbols

15 files changed, 320 insertions, 297 deletions

diff --git a/cmd/sbc_harness/config/config.h b/cmd/sbc_harness/config/config.h
index fd089b1..da3edad 100644
--- a/cmd/sbc_harness/config/config.h
+++ b/cmd/sbc_harness/config/config.h
@@ -32,6 +32,8 @@
 
 /* 9P *************************************************************************/
 
+#define CONFIG_9P_MAX_ERR_SIZE          128 /* 128 is what Plan 9 4e uses */
+
 /**
  * This max-msg-size is sized so that a Twrite message can return
  * 8KiB of data.
@@ -49,22 +51,22 @@
  * negotiated.  In Plan 9 1e it was (8*1024)+128, and was bumped to
  * (8*1024)+160 in 2e and 3e.
  */
-#define CONFIG_9P_MAX_MSG_SIZE        ((4*1024)+24)
+#define CONFIG_9P_SRV_MAX_MSG_SIZE      ((4*1024)+24)
 /**
  * Maximum host-data-structure size.  A message may be larger in
  * unmarshaled-host-structures than marshaled-net-bytes due to (1)
- * struct padding, (2) nul-terminator byes for strings.
+ * struct padding, (2) array pointers.
  */
-#define CONFIG_9P_MAX_HOSTMSG_SIZE    CONFIG_9P_MAX_MSG_SIZE+16
-#define CONFIG_9P_MAX_ERR_SIZE        128 /* 128 is what Plan 9 4e uses */
-#define CONFIG_9P_SRV_MAX_FIDS        16
-#define CONFIG_9P_SRV_MAX_REQS        2
-#define CONFIG_9P_SRV_MAX_DEPTH       3
-#define CONFIG_9P_ENABLE_9P2000       1 /* bool */
-#define CONFIG_9P_ENABLE_9P2000_u     1 /* bool */
-#define CONFIG_9P_ENABLE_9P2000_e     0 /* bool */
-#define CONFIG_9P_ENABLE_9P2000_L     0 /* bool */
-#define CONFIG_9P_ENABLE_9P2000_p9p   0 /* bool */
+#define CONFIG_9P_SRV_MAX_HOSTMSG_SIZE  CONFIG_9P_SRV_MAX_MSG_SIZE+16
+#define CONFIG_9P_SRV_MAX_FIDS          16
+#define CONFIG_9P_SRV_MAX_REQS          2
+#define CONFIG_9P_SRV_MAX_DEPTH         3
+
+#define CONFIG_9P_ENABLE_9P2000         1 /* bool */
+#define CONFIG_9P_ENABLE_9P2000_u       1 /* bool */
+#define CONFIG_9P_ENABLE_9P2000_e       0 /* bool */
+#define CONFIG_9P_ENABLE_9P2000_L       0 /* bool */
+#define CONFIG_9P_ENABLE_9P2000_p9p     0 /* bool */
 
 /* DHCP ***********************************************************************/
 
diff --git a/lib9p/9p.c b/lib9p/9p.c
index c11bcd6..e90ef30 100644
--- a/lib9p/9p.c
+++ b/lib9p/9p.c
@@ -9,32 +9,10 @@
 #include <stdio.h>    /* for vsnprintf() */
 #include <string.h>   /* for strncpy() */
 
-#define LOG_NAME 9P
-#include <libmisc/log.h> /* for const_byte_str() */
-
 #include <lib9p/9p.h>
 
-#include "internal.h"
-
 /* strings ********************************************************************/
 
-const char *lib9p_version_str(enum lib9p_version ver) {
-#pragma GCC diagnostic push
-#pragma GCC diagnostic ignored "-Wtype-limits"
-	assert(0 <= ver && ver < LIB9P_VER_NUM);
-#pragma GCC diagnostic pop
-	return _lib9p_table_ver_name[ver];
-}
-
-const char *lib9p_msgtype_str(enum lib9p_version ver, enum lib9p_msg_type typ) {
-#pragma GCC diagnostic push
-#pragma GCC diagnostic ignored "-Wtype-limits"
-	assert(0 <= ver && ver < LIB9P_VER_NUM);
-	assert(0 <= typ && typ <= 0xFF);
-#pragma GCC diagnostic pop
-	return _lib9p_table_msg_name[ver][typ] ?: const_byte_str(typ);
-}
-
 struct lib9p_s lib9p_str(char *s) {
 	if (!s)
 		return (struct lib9p_s){0};
@@ -114,131 +92,3 @@ int lib9p_errorf(struct lib9p_ctx *ctx, lib9p_errno_t linux_errno, char const *f
 
 	return -1;
 }
-
-/* main message functions *****************************************************/
-
-static
-ssize_t _lib9p_validate(uint8_t                          xxx_low_typ_bit,
-                        const char                      *xxx_errmsg,
-                        const struct _lib9p_recv_tentry  xxx_table[LIB9P_VER_NUM][0x80],
-                        struct lib9p_ctx *ctx, uint8_t *net_bytes) {
-	/* Inspect the first 5 bytes ourselves.  */
-	uint32_t net_size = uint32le_decode(net_bytes);
-	if (net_size < 5)
-		return lib9p_error(ctx, LINUX_EBADMSG, "message is impossibly short");
-	uint8_t typ = net_bytes[4];
-	if (typ % 2 != xxx_low_typ_bit)
-		return lib9p_errorf(ctx, LINUX_EOPNOTSUPP, "%s: message_type=%s", xxx_errmsg,
-		                    lib9p_msgtype_str(ctx->version, typ));
-	struct _lib9p_recv_tentry tentry = xxx_table[ctx->version][typ/2];
-	if (!tentry.validate)
-		return lib9p_errorf(ctx, LINUX_EOPNOTSUPP, "unknown message type: %s (protocol_version=%s)",
-		                    lib9p_msgtype_str(ctx->version, typ), lib9p_version_str(ctx->version));
-
-	/* Now use the message-type-specific tentry to process the whole thing.  */
-	return tentry.validate(ctx, net_size, net_bytes);
-}
-
-ssize_t lib9p_Tmsg_validate(struct lib9p_ctx *ctx, uint8_t *net_bytes) {
-	return _lib9p_validate(0, "expected a T-message but got an R-message", _lib9p_table_Tmsg_recv,
-	                       ctx, net_bytes);
-}
-
-ssize_t lib9p_Rmsg_validate(struct lib9p_ctx *ctx, uint8_t *net_bytes) {
-	return _lib9p_validate(1, "expected an R-message but got a T-message", _lib9p_table_Rmsg_recv,
-	                       ctx, net_bytes);
-}
-
-static
-void _lib9p_unmarshal(const struct _lib9p_recv_tentry xxx_table[LIB9P_VER_NUM][0x80],
-                      struct lib9p_ctx *ctx, uint8_t *net_bytes,
-                      enum lib9p_msg_type *ret_typ, void *ret_body) {
-	enum lib9p_msg_type typ = net_bytes[4];
-	*ret_typ = typ;
-	struct _lib9p_recv_tentry tentry = xxx_table[ctx->version][typ/2];
-
-	tentry.unmarshal(ctx, net_bytes, ret_body);
-}
-
-void lib9p_Tmsg_unmarshal(struct lib9p_ctx *ctx, uint8_t *net_bytes,
-                          enum lib9p_msg_type *ret_typ, void *ret_body) {
-	_lib9p_unmarshal(_lib9p_table_Tmsg_recv,
-	                 ctx, net_bytes, ret_typ, ret_body);
-}
-
-void lib9p_Rmsg_unmarshal(struct lib9p_ctx *ctx, uint8_t *net_bytes,
-                          enum lib9p_msg_type *ret_typ, void *ret_body) {
-	_lib9p_unmarshal(_lib9p_table_Rmsg_recv,
-	                 ctx, net_bytes, ret_typ, ret_body);
-}
-
-static
-bool _lib9p_marshal(const struct _lib9p_send_tentry xxx_table[LIB9P_VER_NUM][0x80],
-                    struct lib9p_ctx *ctx, enum lib9p_msg_type typ, void *body,
-                    size_t *ret_iov_cnt, struct iovec *ret_iov, uint8_t *ret_copied) {
-	struct _marshal_ret ret = {
-		.net_iov_cnt     = 1,
-		.net_iov         = ret_iov,
-		.net_copied_size = 0,
-		.net_copied      = ret_copied,
-	};
-
-	struct _lib9p_send_tentry tentry = xxx_table[ctx->version][typ/2];
-	bool ret_erred = tentry.marshal(ctx, body, &ret);
-	if (ret_iov[ret.net_iov_cnt-1].iov_len == 0)
-		ret.net_iov_cnt--;
-	*ret_iov_cnt = ret.net_iov_cnt;
-	return ret_erred;
-}
-
-bool lib9p_Tmsg_marshal(struct lib9p_ctx *ctx, enum lib9p_msg_type typ, void *body,
-                        struct lib9p_Tmsg_send_buf *ret) {
-	assert(typ % 2 == 0);
-	memset(ret, 0, sizeof(*ret));
-	return _lib9p_marshal(_lib9p_table_Tmsg_send,
-	                      ctx, typ, body,
-	                      &ret->iov_cnt, ret->iov, ret->copied);
-}
-
-bool lib9p_Rmsg_marshal(struct lib9p_ctx *ctx, enum lib9p_msg_type typ, void *body,
-                        struct lib9p_Rmsg_send_buf *ret) {
-	assert(typ % 2 == 1);
-	memset(ret, 0, sizeof(*ret));
-	return _lib9p_marshal(_lib9p_table_Rmsg_send,
-	                      ctx, typ, body,
-	                      &ret->iov_cnt, ret->iov, ret->copied);
-}
-
-/* `struct lib9p_stat` helpers ************************************************/
-
-bool lib9p_stat_validate(struct lib9p_ctx *ctx, uint32_t net_size, uint8_t *net_bytes,
-                         uint32_t *ret_net_size, ssize_t *ret_host_size) {
-	ssize_t host_size = _lib9p_stat_validate(ctx, net_size, net_bytes, ret_net_size);
-	if (host_size < 0)
-		return true;
-	if (ret_host_size)
-		*ret_host_size = host_size;
-	return false;
-}
-
-void lib9p_stat_unmarshal(struct lib9p_ctx *ctx, uint8_t *net_bytes,
-                          struct lib9p_stat *ret) {
-	_lib9p_stat_unmarshal(ctx, net_bytes, ret);
-}
-
-uint32_t lib9p_stat_marshal(struct lib9p_ctx *ctx, uint32_t max_net_size, struct lib9p_stat *obj,
-                            uint8_t *ret_bytes) {
-	struct lib9p_ctx _ctx = *ctx;
-	_ctx.max_msg_size = max_net_size;
-
-	struct iovec iov = {0};
-	struct _marshal_ret ret = {
-		.net_iov_cnt     = 1,
-		.net_iov         = &iov,
-		.net_copied_size = 0,
-		.net_copied      = ret_bytes,
-	};
-	if (_lib9p_stat_marshal(&_ctx, obj, &ret))
-		return 0;
-	return ret.net_iov[0].iov_len;
-}
diff --git a/lib9p/9p.generated.c b/lib9p/9p.generated.c
index 3031392..2549b87 100644
--- a/lib9p/9p.generated.c
+++ b/lib9p/9p.generated.c
@@ -6,10 +6,12 @@
 #include <string.h>   /* for memset() */
 
 #include <libmisc/assert.h>
+#include <libmisc/endian.h>
 
 #include <lib9p/9p.h>
 
-#include "internal.h"
+#include "tables.h"
+#include "utf8.h"
 
 /* utilities ******************************************************************/
 #if CONFIG_9P_ENABLE_9P2000
diff --git a/lib9p/CMakeLists.txt b/lib9p/CMakeLists.txt
index 707308e..9240aa3 100644
--- a/lib9p/CMakeLists.txt
+++ b/lib9p/CMakeLists.txt
@@ -8,6 +8,7 @@ target_include_directories(lib9p SYSTEM INTERFACE ${CMAKE_CURRENT_SOURCE_DIR}/in
 target_sources(lib9p INTERFACE
 	9p.generated.c
 	9p.c
+	tables.c
 	srv.c
 )
 target_link_libraries(lib9p INTERFACE
diff --git a/lib9p/include/lib9p/9p.h b/lib9p/include/lib9p/9p.h
index ffac453..4cdb997 100644
--- a/lib9p/include/lib9p/9p.h
+++ b/lib9p/include/lib9p/9p.h
@@ -18,6 +18,7 @@
 #ifndef CONFIG_9P_MAX_ERR_SIZE
 	#error config.h must define CONFIG_9P_MAX_ERR_SIZE
 #endif
+static_assert(CONFIG_9P_MAX_ERR_SIZE <= UINT16_MAX);
 
 /* constants ******************************************************************/
 
@@ -60,6 +61,10 @@ int lib9p_error(struct lib9p_ctx *ctx, lib9p_errno_t linux_errno, char const *ms
 /** Write a printf-style error into ctx, return -1.  */
 int lib9p_errorf(struct lib9p_ctx *ctx, lib9p_errno_t linux_errno, char const *fmt, ...) [[gnu::format(printf, 3, 4)]];
 
+/* misc utilities *************************************************************/
+
+uint32_t lib9p_version_min_msg_size(enum lib9p_version);
+
 /* main T-message functions ***************************************************/
 
 /**
diff --git a/lib9p/internal.h b/lib9p/internal.h
deleted file mode 100644
index 51bf792..0000000
--- a/lib9p/internal.h
+++ /dev/null
@@ -1,99 +0,0 @@
-/* lib9p/internal.h - Internal machinery shared between parts of lib9p
- *
- * Copyright (C) 2024-2025  Luke T. Shumaker <lukeshu@lukeshu.com>
- * SPDX-License-Identifier: AGPL-3.0-or-later
- */
-
-#ifndef _LIB9P_INTERNAL_H_
-#define _LIB9P_INTERNAL_H_
-
-#include <stddef.h> /* for size_t */
-#include <limits.h> /* for SSIZE_MAX, not set by newlib */
-#ifndef SSIZE_MAX
-#define SSIZE_MAX (SIZE_MAX >> 1)
-#endif
-
-#include <libmisc/endian.h>
-#include <libmisc/macro.h>
-
-#include <lib9p/9p.h>
-
-/* configuration **************************************************************/
-
-#include "config.h"
-
-#ifndef CONFIG_9P_MAX_MSG_SIZE
-	#error config.h must define CONFIG_9P_MAX_MSG_SIZE
-#endif
-#ifndef CONFIG_9P_MAX_HOSTMSG_SIZE
-	#error config.h must define CONFIG_9P_MAX_HOSTMSG_SIZE
-#endif
-#ifndef CONFIG_9P_MAX_ERR_SIZE
-	#error config.h must define CONFIG_9P_MAX_ERR_SIZE
-#endif
-
-static_assert(CONFIG_9P_MAX_ERR_SIZE <= UINT16_MAX);
-static_assert(CONFIG_9P_MAX_MSG_SIZE <= CONFIG_9P_MAX_HOSTMSG_SIZE);
-static_assert(CONFIG_9P_MAX_HOSTMSG_SIZE <= SSIZE_MAX);
-
-/* tables / exports ***********************************************************/
-
-typedef ssize_t (*_validate_fn_t)(struct lib9p_ctx *ctx, uint32_t net_size, uint8_t *net_bytes);
-typedef void (*_unmarshal_fn_t)(struct lib9p_ctx *ctx, uint8_t *net_bytes, void *out);
-
-struct _marshal_ret {
-	size_t                   net_iov_cnt;
-	struct iovec            *net_iov;
-	size_t                   net_copied_size;
-	uint8_t                 *net_copied;
-};
-typedef bool (*_marshal_fn_t)(struct lib9p_ctx *ctx, void *host_val, struct _marshal_ret *ret);
-
-struct _lib9p_recv_tentry {
-	_validate_fn_t          validate;
-	_unmarshal_fn_t         unmarshal;
-};
-
-struct _lib9p_send_tentry {
-	_marshal_fn_t           marshal;
-};
-
-extern const char *const _lib9p_table_ver_name[LIB9P_VER_NUM];
-extern const char *const _lib9p_table_msg_name[LIB9P_VER_NUM][0x100];
-extern const uint32_t _lib9p_table_msg_min_size[LIB9P_VER_NUM];
-extern const struct _lib9p_recv_tentry _lib9p_table_Tmsg_recv[LIB9P_VER_NUM][0x80];
-extern const struct _lib9p_recv_tentry _lib9p_table_Rmsg_recv[LIB9P_VER_NUM][0x80];
-extern const struct _lib9p_send_tentry _lib9p_table_Tmsg_send[LIB9P_VER_NUM][0x80];
-extern const struct _lib9p_send_tentry _lib9p_table_Rmsg_send[LIB9P_VER_NUM][0x80];
-
-ssize_t _lib9p_stat_validate(struct lib9p_ctx *ctx, uint32_t net_size, uint8_t *net_bytes, uint32_t *ret_net_size);
-void _lib9p_stat_unmarshal(struct lib9p_ctx *ctx, uint8_t *net_bytes, void *out);
-bool _lib9p_stat_marshal(struct lib9p_ctx *ctx, struct lib9p_stat *val, struct _marshal_ret *ret);
-
-/* unmarshal utilities ********************************************************/
-
-static inline bool _is_valid_utf8(uint8_t *str, size_t len, bool forbid_nul) {
-	uint32_t ch;
-	uint8_t chlen;
-	assert(str);
-	for (size_t pos = 0; pos < len;) {
-		if      ((str[pos] & 0b10000000) == 0b00000000) { ch = str[pos] & 0b01111111; chlen = 1; }
-		else if ((str[pos] & 0b11100000) == 0b11000000) { ch = str[pos] & 0b00011111; chlen = 2; }
-		else if ((str[pos] & 0b11110000) == 0b11100000) { ch = str[pos] & 0b00001111; chlen = 3; }
-		else if ((str[pos] & 0b11111000) == 0b11110000) { ch = str[pos] & 0b00000111; chlen = 4; }
-		else return false;
-		if ((ch == 0 && (chlen != 1 || forbid_nul)) || pos + chlen > len) return false;
-		for (uint8_t i = 1; i < chlen; i++) {
-			if ((str[pos+i] & 0b11000000) != 0b10000000) return false;
-			ch = (ch << 6) | (str[pos+i] & 0b00111111);
-		}
-		if (ch > 0x10FFFF) return false;
-		pos += chlen;
-	}
-	return true;
-}
-
-#define is_valid_utf8(str, len)                 _is_valid_utf8(str, len, false)
-#define is_valid_utf8_without_nul(str, len)     _is_valid_utf8(str, len, true)
-
-#endif /* _LIB9P_INTERNAL_H_ */
diff --git a/lib9p/map.h b/lib9p/map.h
index ab9564f..c5eab0f 100644
--- a/lib9p/map.h
+++ b/lib9p/map.h
@@ -4,8 +4,6 @@
  * SPDX-License-Identifier: AGPL-3.0-or-later
  */
 
-#include "internal.h"
-
 /**
  * `#define` `NAME`, `KEY_T`, `VAL_T`, and `CAP`; then `#include
  * "map.h".
diff --git a/lib9p/protogen/c.py b/lib9p/protogen/c.py
index 0c9b21f..c487c03 100644
--- a/lib9p/protogen/c.py
+++ b/lib9p/protogen/c.py
@@ -29,10 +29,12 @@ def gen_c(versions: set[str], typs: list[idl.UserType]) -> str:
 #include <string.h>   /* for memset() */
 
 #include <libmisc/assert.h>
+#include <libmisc/endian.h>
 
 #include <lib9p/9p.h>
 
-#include "internal.h"
+#include "tables.h"
+#include "utf8.h"
 """
 
     # utilities ################################################################
diff --git a/lib9p/srv.c b/lib9p/srv.c
index a29a4cb..2723207 100644
--- a/lib9p/srv.c
+++ b/lib9p/srv.c
@@ -6,12 +6,18 @@
 
 #include <alloca.h>
 #include <inttypes.h> /* for PRI* */
+#include <stddef.h>   /* for size_t */
+#include <limits.h>   /* for SSIZE_MAX, not set by newlib */
+#ifndef SSIZE_MAX
+#define SSIZE_MAX (SIZE_MAX >> 1)
+#endif
 
 #include <libcr/coroutine.h>
 #include <libcr_ipc/chan.h>
 #include <libcr_ipc/mutex.h>
 #include <libcr_ipc/select.h>
 #include <libmisc/assert.h>
+#include <libmisc/endian.h>
 #include <libhw/generic/net.h>
 
 #define LOG_NAME 9P_SRV
@@ -20,7 +26,9 @@
 #define IMPLEMENTATION_FOR_LIB9P_SRV_H YES
 #include <lib9p/srv.h>
 
-#include "internal.h"
+/* config *********************************************************************/
+
+#include "config.h"
 
 #ifndef CONFIG_9P_SRV_MAX_FIDS
 	#error config.h must define CONFIG_9P_SRV_MAX_FIDS
@@ -32,6 +40,14 @@
 	/* 1=just the root dir, 2=just files in the root dir, 3=1 subdir, ... */
 	#error config.h must define CONFIG_9P_SRV_MAX_DEPTH
 #endif
+#ifndef CONFIG_9P_SRV_MAX_MSG_SIZE
+	#error config.h must define CONFIG_9P_SRV_MAX_MSG_SIZE
+#endif
+#ifndef CONFIG_9P_SRV_MAX_HOSTMSG_SIZE
+	#error config.h must define CONFIG_9P_SRV_MAX_HOSTMSG_SIZE
+#endif
+static_assert(CONFIG_9P_SRV_MAX_MSG_SIZE <= CONFIG_9P_SRV_MAX_HOSTMSG_SIZE);
+static_assert(CONFIG_9P_SRV_MAX_HOSTMSG_SIZE <= SSIZE_MAX);
 
 /* context ********************************************************************/
 
@@ -218,7 +234,7 @@ static void handle_message(struct _lib9p_srv_req *ctx);
 
 	srv->readers++;
 
-	uint32_t initial_rerror_overhead = _lib9p_table_msg_min_size[LIB9P_VER_unknown];
+	uint32_t initial_rerror_overhead = lib9p_version_min_msg_size(LIB9P_VER_unknown);
 
 	for (;;) {
 		struct _srv_conn conn = {
@@ -238,7 +254,7 @@ static void handle_message(struct _lib9p_srv_req *ctx);
 		struct _srv_sess sess = {
 			.parent_conn     = &conn,
 			.version         = LIB9P_VER_unknown,
-			.max_msg_size    = CONFIG_9P_MAX_MSG_SIZE,
+			.max_msg_size    = CONFIG_9P_SRV_MAX_MSG_SIZE,
 			.rerror_overhead = initial_rerror_overhead,
 			.initialized     = false,
 		};
@@ -397,7 +413,7 @@ static tmessage_handler tmessage_handlers[0x100] = {
 
 static void handle_message(struct _lib9p_srv_req *ctx) {
 	uint8_t *host_req = NULL;
-	uint8_t host_resp[CONFIG_9P_MAX_HOSTMSG_SIZE];
+	uint8_t host_resp[CONFIG_9P_SRV_MAX_HOSTMSG_SIZE];
 
 	/* Unmarshal it.  */
 	ssize_t host_size = lib9p_Tmsg_validate(&ctx->ctx.basectx, ctx->net_bytes);
@@ -573,7 +589,7 @@ static void handle_Tversion(struct _lib9p_srv_req *ctx,
 #endif
 	}
 
-	uint32_t min_msg_size = _lib9p_table_msg_min_size[version];
+	uint32_t min_msg_size = lib9p_version_min_msg_size(version);
 	if (req->max_msg_size < min_msg_size) {
 		lib9p_errorf(&ctx->ctx.basectx,
 		             LINUX_EDOM, "requested max_msg_size is less than minimum for %s (%"PRIu32" < %"PRIu32")",
@@ -582,8 +598,8 @@ static void handle_Tversion(struct _lib9p_srv_req *ctx,
 	}
 
 	resp->version = lib9p_str((char *)lib9p_version_str(version)); /* cast to discard "const" qualifier */
-	resp->max_msg_size = (CONFIG_9P_MAX_MSG_SIZE < req->max_msg_size)
-		? CONFIG_9P_MAX_MSG_SIZE
+	resp->max_msg_size = (CONFIG_9P_SRV_MAX_MSG_SIZE < req->max_msg_size)
+		? CONFIG_9P_SRV_MAX_MSG_SIZE
 		: req->max_msg_size;
 
 	/* Close the old session.  */
diff --git a/lib9p/tables.c b/lib9p/tables.c
new file mode 100644
index 0000000..34ac1af
--- /dev/null
+++ b/lib9p/tables.c
@@ -0,0 +1,165 @@
+/* lib9p/tables.c - Access tables of version and message information
+ *
+ * Copyright (C) 2024-2025  Luke T. Shumaker <lukeshu@lukeshu.com>
+ * SPDX-License-Identifier: AGPL-3.0-or-later
+ */
+
+#include <string.h>
+
+#include <libmisc/endian.h>
+#include <libmisc/log.h> /* for const_byte_str() */
+
+#include "tables.h"
+
+const char *lib9p_version_str(enum lib9p_version ver) {
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wtype-limits"
+	assert(0 <= ver && ver < LIB9P_VER_NUM);
+#pragma GCC diagnostic pop
+	return _lib9p_table_ver_name[ver];
+}
+
+uint32_t lib9p_version_min_msg_size(enum lib9p_version ver) {
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wtype-limits"
+	assert(0 <= ver && ver < LIB9P_VER_NUM);
+#pragma GCC diagnostic pop
+	return _lib9p_table_msg_min_size[ver];
+}
+
+const char *lib9p_msgtype_str(enum lib9p_version ver, enum lib9p_msg_type typ) {
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wtype-limits"
+	assert(0 <= ver && ver < LIB9P_VER_NUM);
+	assert(0 <= typ && typ <= 0xFF);
+#pragma GCC diagnostic pop
+	return _lib9p_table_msg_name[ver][typ] ?: const_byte_str(typ);
+}
+
+/* main message functions *****************************************************/
+
+static
+ssize_t _lib9p_validate(uint8_t                          xxx_low_typ_bit,
+                        const char                      *xxx_errmsg,
+                        const struct _lib9p_recv_tentry  xxx_table[LIB9P_VER_NUM][0x80],
+                        struct lib9p_ctx *ctx, uint8_t *net_bytes) {
+	/* Inspect the first 5 bytes ourselves.  */
+	uint32_t net_size = uint32le_decode(net_bytes);
+	if (net_size < 5)
+		return lib9p_error(ctx, LINUX_EBADMSG, "message is impossibly short");
+	uint8_t typ = net_bytes[4];
+	if (typ % 2 != xxx_low_typ_bit)
+		return lib9p_errorf(ctx, LINUX_EOPNOTSUPP, "%s: message_type=%s", xxx_errmsg,
+		                    lib9p_msgtype_str(ctx->version, typ));
+	struct _lib9p_recv_tentry tentry = xxx_table[ctx->version][typ/2];
+	if (!tentry.validate)
+		return lib9p_errorf(ctx, LINUX_EOPNOTSUPP, "unknown message type: %s (protocol_version=%s)",
+		                    lib9p_msgtype_str(ctx->version, typ), lib9p_version_str(ctx->version));
+
+	/* Now use the message-type-specific tentry to process the whole thing.  */
+	return tentry.validate(ctx, net_size, net_bytes);
+}
+
+ssize_t lib9p_Tmsg_validate(struct lib9p_ctx *ctx, uint8_t *net_bytes) {
+	return _lib9p_validate(0, "expected a T-message but got an R-message", _lib9p_table_Tmsg_recv,
+	                       ctx, net_bytes);
+}
+
+ssize_t lib9p_Rmsg_validate(struct lib9p_ctx *ctx, uint8_t *net_bytes) {
+	return _lib9p_validate(1, "expected an R-message but got a T-message", _lib9p_table_Rmsg_recv,
+	                       ctx, net_bytes);
+}
+
+static
+void _lib9p_unmarshal(const struct _lib9p_recv_tentry xxx_table[LIB9P_VER_NUM][0x80],
+                      struct lib9p_ctx *ctx, uint8_t *net_bytes,
+                      enum lib9p_msg_type *ret_typ, void *ret_body) {
+	enum lib9p_msg_type typ = net_bytes[4];
+	*ret_typ = typ;
+	struct _lib9p_recv_tentry tentry = xxx_table[ctx->version][typ/2];
+
+	tentry.unmarshal(ctx, net_bytes, ret_body);
+}
+
+void lib9p_Tmsg_unmarshal(struct lib9p_ctx *ctx, uint8_t *net_bytes,
+                          enum lib9p_msg_type *ret_typ, void *ret_body) {
+	_lib9p_unmarshal(_lib9p_table_Tmsg_recv,
+	                 ctx, net_bytes, ret_typ, ret_body);
+}
+
+void lib9p_Rmsg_unmarshal(struct lib9p_ctx *ctx, uint8_t *net_bytes,
+                          enum lib9p_msg_type *ret_typ, void *ret_body) {
+	_lib9p_unmarshal(_lib9p_table_Rmsg_recv,
+	                 ctx, net_bytes, ret_typ, ret_body);
+}
+
+static
+bool _lib9p_marshal(const struct _lib9p_send_tentry xxx_table[LIB9P_VER_NUM][0x80],
+                    struct lib9p_ctx *ctx, enum lib9p_msg_type typ, void *body,
+                    size_t *ret_iov_cnt, struct iovec *ret_iov, uint8_t *ret_copied) {
+	struct _marshal_ret ret = {
+		.net_iov_cnt     = 1,
+		.net_iov         = ret_iov,
+		.net_copied_size = 0,
+		.net_copied      = ret_copied,
+	};
+
+	struct _lib9p_send_tentry tentry = xxx_table[ctx->version][typ/2];
+	bool ret_erred = tentry.marshal(ctx, body, &ret);
+	if (ret_iov[ret.net_iov_cnt-1].iov_len == 0)
+		ret.net_iov_cnt--;
+	*ret_iov_cnt = ret.net_iov_cnt;
+	return ret_erred;
+}
+
+bool lib9p_Tmsg_marshal(struct lib9p_ctx *ctx, enum lib9p_msg_type typ, void *body,
+                        struct lib9p_Tmsg_send_buf *ret) {
+	assert(typ % 2 == 0);
+	memset(ret, 0, sizeof(*ret));
+	return _lib9p_marshal(_lib9p_table_Tmsg_send,
+	                      ctx, typ, body,
+	                      &ret->iov_cnt, ret->iov, ret->copied);
+}
+
+bool lib9p_Rmsg_marshal(struct lib9p_ctx *ctx, enum lib9p_msg_type typ, void *body,
+                        struct lib9p_Rmsg_send_buf *ret) {
+	assert(typ % 2 == 1);
+	memset(ret, 0, sizeof(*ret));
+	return _lib9p_marshal(_lib9p_table_Rmsg_send,
+	                      ctx, typ, body,
+	                      &ret->iov_cnt, ret->iov, ret->copied);
+}
+
+/* `struct lib9p_stat` helpers ************************************************/
+
+bool lib9p_stat_validate(struct lib9p_ctx *ctx, uint32_t net_size, uint8_t *net_bytes,
+                         uint32_t *ret_net_size, ssize_t *ret_host_size) {
+	ssize_t host_size = _lib9p_stat_validate(ctx, net_size, net_bytes, ret_net_size);
+	if (host_size < 0)
+		return true;
+	if (ret_host_size)
+		*ret_host_size = host_size;
+	return false;
+}
+
+void lib9p_stat_unmarshal(struct lib9p_ctx *ctx, uint8_t *net_bytes,
+                          struct lib9p_stat *ret) {
+	_lib9p_stat_unmarshal(ctx, net_bytes, ret);
+}
+
+uint32_t lib9p_stat_marshal(struct lib9p_ctx *ctx, uint32_t max_net_size, struct lib9p_stat *obj,
+                            uint8_t *ret_bytes) {
+	struct lib9p_ctx _ctx = *ctx;
+	_ctx.max_msg_size = max_net_size;
+
+	struct iovec iov = {0};
+	struct _marshal_ret ret = {
+		.net_iov_cnt     = 1,
+		.net_iov         = &iov,
+		.net_copied_size = 0,
+		.net_copied      = ret_bytes,
+	};
+	if (_lib9p_stat_marshal(&_ctx, obj, &ret))
+		return 0;
+	return ret.net_iov[0].iov_len;
+}
diff --git a/lib9p/tables.h b/lib9p/tables.h
new file mode 100644
index 0000000..5591e4b
--- /dev/null
+++ b/lib9p/tables.h
@@ -0,0 +1,44 @@
+/* lib9p/tables.h - Declare tables of version and message information
+ *
+ * Copyright (C) 2024-2025  Luke T. Shumaker <lukeshu@lukeshu.com>
+ * SPDX-License-Identifier: AGPL-3.0-or-later
+ */
+
+#ifndef _LIB9P_TABLES_H_
+#define _LIB9P_TABLES_H_
+
+#include <lib9p/9p.h>
+
+typedef ssize_t (*_validate_fn_t)(struct lib9p_ctx *ctx, uint32_t net_size, uint8_t *net_bytes);
+typedef void (*_unmarshal_fn_t)(struct lib9p_ctx *ctx, uint8_t *net_bytes, void *out);
+
+struct _marshal_ret {
+	size_t                   net_iov_cnt;
+	struct iovec            *net_iov;
+	size_t                   net_copied_size;
+	uint8_t                 *net_copied;
+};
+typedef bool (*_marshal_fn_t)(struct lib9p_ctx *ctx, void *host_val, struct _marshal_ret *ret);
+
+struct _lib9p_recv_tentry {
+	_validate_fn_t          validate;
+	_unmarshal_fn_t         unmarshal;
+};
+
+struct _lib9p_send_tentry {
+	_marshal_fn_t           marshal;
+};
+
+extern const char *const _lib9p_table_ver_name[LIB9P_VER_NUM];
+extern const char *const _lib9p_table_msg_name[LIB9P_VER_NUM][0x100];
+extern const uint32_t _lib9p_table_msg_min_size[LIB9P_VER_NUM];
+extern const struct _lib9p_recv_tentry _lib9p_table_Tmsg_recv[LIB9P_VER_NUM][0x80];
+extern const struct _lib9p_recv_tentry _lib9p_table_Rmsg_recv[LIB9P_VER_NUM][0x80];
+extern const struct _lib9p_send_tentry _lib9p_table_Tmsg_send[LIB9P_VER_NUM][0x80];
+extern const struct _lib9p_send_tentry _lib9p_table_Rmsg_send[LIB9P_VER_NUM][0x80];
+
+ssize_t _lib9p_stat_validate(struct lib9p_ctx *ctx, uint32_t net_size, uint8_t *net_bytes, uint32_t *ret_net_size);
+void _lib9p_stat_unmarshal(struct lib9p_ctx *ctx, uint8_t *net_bytes, void *out);
+bool _lib9p_stat_marshal(struct lib9p_ctx *ctx, struct lib9p_stat *val, struct _marshal_ret *ret);
+
+#endif /* _LIB9P_TABLES_H_ */
diff --git a/lib9p/tests/test_compile_config/config.h b/lib9p/tests/test_compile_config/config.h
index 38ab0c0..cc8eec1 100644
--- a/lib9p/tests/test_compile_config/config.h
+++ b/lib9p/tests/test_compile_config/config.h
@@ -9,20 +9,20 @@
 
 /* 9P *************************************************************************/
 
-#define CONFIG_9P_MAX_MSG_SIZE        ((4*1024)+24)
-#define CONFIG_9P_MAX_HOSTMSG_SIZE    CONFIG_9P_MAX_MSG_SIZE+16
-#define CONFIG_9P_MAX_ERR_SIZE        128
-#define CONFIG_9P_MAX_9P2000_e_WELEM  16
-
-#define CONFIG_9P_SRV_MAX_FIDS        16
-#define CONFIG_9P_SRV_MAX_REQS        2
-#define CONFIG_9P_SRV_MAX_DEPTH       3
-
-#define CONFIG_9P_ENABLE_9P2000       1 /* bool */
-#define CONFIG_9P_ENABLE_9P2000_u     1 /* bool */
-#define CONFIG_9P_ENABLE_9P2000_e     1 /* bool */
-#define CONFIG_9P_ENABLE_9P2000_L     1 /* bool */
-#define CONFIG_9P_ENABLE_9P2000_p9p   1 /* bool */
+#define CONFIG_9P_MAX_ERR_SIZE          128
+#define CONFIG_9P_MAX_9P2000_e_WELEM    16
+
+#define CONFIG_9P_SRV_MAX_MSG_SIZE      ((4*1024)+24)
+#define CONFIG_9P_SRV_MAX_HOSTMSG_SIZE  CONFIG_9P_SRV_MAX_MSG_SIZE+16
+#define CONFIG_9P_SRV_MAX_FIDS          16
+#define CONFIG_9P_SRV_MAX_REQS          2
+#define CONFIG_9P_SRV_MAX_DEPTH         3
+
+#define CONFIG_9P_ENABLE_9P2000         1 /* bool */
+#define CONFIG_9P_ENABLE_9P2000_u       1 /* bool */
+#define CONFIG_9P_ENABLE_9P2000_e       1 /* bool */
+#define CONFIG_9P_ENABLE_9P2000_L       1 /* bool */
+#define CONFIG_9P_ENABLE_9P2000_p9p     1 /* bool */
 
 /* COROUTINE ******************************************************************/
 
diff --git a/lib9p/tests/test_server/config/config.h b/lib9p/tests/test_server/config/config.h
index afe6dc0..03143e1 100644
--- a/lib9p/tests/test_server/config/config.h
+++ b/lib9p/tests/test_server/config/config.h
@@ -12,6 +12,8 @@
 
 /* 9P *************************************************************************/
 
+#define CONFIG_9P_MAX_ERR_SIZE          128 /* 128 is what Plan 9 4e uses */
+
 /**
  * This max-msg-size is sized so that a Twrite message can return
  * 8KiB of data.
@@ -29,22 +31,22 @@
  * negotiated.  In Plan 9 1e it was (8*1024)+128, and was bumped to
  * (8*1024)+160 in 2e and 3e.
  */
-#define CONFIG_9P_MAX_MSG_SIZE        ((4*1024)+24)
+#define CONFIG_9P_SRV_MAX_MSG_SIZE      ((4*1024)+24)
 /**
  * Maximum host-data-structure size.  A message may be larger in
  * unmarshaled-host-structures than marshaled-net-bytes due to (1)
  * struct padding, (2) array pointers.
  */
-#define CONFIG_9P_MAX_HOSTMSG_SIZE    CONFIG_9P_MAX_MSG_SIZE+16
-#define CONFIG_9P_MAX_ERR_SIZE        128 /* 128 is what Plan 9 4e uses */
-#define CONFIG_9P_SRV_MAX_FIDS        16
-#define CONFIG_9P_SRV_MAX_REQS        2
-#define CONFIG_9P_SRV_MAX_DEPTH       3
-#define CONFIG_9P_ENABLE_9P2000       1 /* bool */
-#define CONFIG_9P_ENABLE_9P2000_u     1 /* bool */
-#define CONFIG_9P_ENABLE_9P2000_e     0 /* bool */
-#define CONFIG_9P_ENABLE_9P2000_L     0 /* bool */
-#define CONFIG_9P_ENABLE_9P2000_p9p   0 /* bool */
+#define CONFIG_9P_SRV_MAX_HOSTMSG_SIZE  CONFIG_9P_SRV_MAX_MSG_SIZE+16
+#define CONFIG_9P_SRV_MAX_FIDS          16
+#define CONFIG_9P_SRV_MAX_REQS          2
+#define CONFIG_9P_SRV_MAX_DEPTH         3
+
+#define CONFIG_9P_ENABLE_9P2000         1 /* bool */
+#define CONFIG_9P_ENABLE_9P2000_u       1 /* bool */
+#define CONFIG_9P_ENABLE_9P2000_e       0 /* bool */
+#define CONFIG_9P_ENABLE_9P2000_L       0 /* bool */
+#define CONFIG_9P_ENABLE_9P2000_p9p     0 /* bool */
 
 /* COROUTINE ******************************************************************/
 
diff --git a/lib9p/utf8.h b/lib9p/utf8.h
new file mode 100644
index 0000000..5ffd674
--- /dev/null
+++ b/lib9p/utf8.h
@@ -0,0 +1,34 @@
+/* lib9p/utf8.h - Internal UTF-8 validation
+ *
+ * Copyright (C) 2024-2025  Luke T. Shumaker <lukeshu@lukeshu.com>
+ * SPDX-License-Identifier: AGPL-3.0-or-later
+ */
+
+#ifndef _LIB9P_UTF8_H_
+#define _LIB9P_UTF8_H_
+
+static inline bool _is_valid_utf8(uint8_t *str, size_t len, bool forbid_nul) {
+	uint32_t ch;
+	uint8_t chlen;
+	assert(str);
+	for (size_t pos = 0; pos < len;) {
+		if      ((str[pos] & 0b10000000) == 0b00000000) { ch = str[pos] & 0b01111111; chlen = 1; }
+		else if ((str[pos] & 0b11100000) == 0b11000000) { ch = str[pos] & 0b00011111; chlen = 2; }
+		else if ((str[pos] & 0b11110000) == 0b11100000) { ch = str[pos] & 0b00001111; chlen = 3; }
+		else if ((str[pos] & 0b11111000) == 0b11110000) { ch = str[pos] & 0b00000111; chlen = 4; }
+		else return false;
+		if ((ch == 0 && (chlen != 1 || forbid_nul)) || pos + chlen > len) return false;
+		for (uint8_t i = 1; i < chlen; i++) {
+			if ((str[pos+i] & 0b11000000) != 0b10000000) return false;
+			ch = (ch << 6) | (str[pos+i] & 0b00111111);
+		}
+		if (ch > 0x10FFFF) return false;
+		pos += chlen;
+	}
+	return true;
+}
+
+#define is_valid_utf8(str, len)                 _is_valid_utf8(str, len, false)
+#define is_valid_utf8_without_nul(str, len)     _is_valid_utf8(str, len, true)
+
+#endif /* _LIB9P_UTF8_H_ */
diff --git a/libmisc/include/libmisc/endian.h b/libmisc/include/libmisc/endian.h
index 665a8c8..75240fe 100644
--- a/libmisc/include/libmisc/endian.h
+++ b/libmisc/include/libmisc/endian.h
@@ -7,6 +7,7 @@
 #ifndef _LIBMISC_ENDIAN_H_
 #define _LIBMISC_ENDIAN_H_
 
+#include <stddef.h> /* for size_t */
 #include <stdint.h> /* for uint{n}_t */
 
 #include <libmisc/assert.h>