Merge branch 'ps/writev' into seen

A compatibility wrapper for writev(3p) has been reintroduced,
including fixes for CMake build and 'MAX_IO_SIZE' limits on NonStop.
Calls to write(3p) in send_sideband() and cat_blob() have been
refactored to use writev(3p) wrappers to reduce syscall overhead.

* ps/writev:
  fast-import: use writev(3p) to send cat-blob responses
  sideband: use writev(3p) to send pktlines
  wrapper: properly handle MAX_IO_SIZE in writev(3p)
  wrapper: introduce writev(3p) wrappers
  compat/posix: introduce writev(3p) wrapper

Makefile

@@ -2045,6 +2045,10 @@ ifdef NO_PREAD
 	COMPAT_CFLAGS += -DNO_PREAD
 	COMPAT_OBJS += compat/pread.o
 endif
+ifdef NO_WRITEV
+	COMPAT_CFLAGS += -DNO_WRITEV
+	COMPAT_OBJS += compat/writev.o
+endif
 ifdef NO_FAST_WORKING_DIRECTORY
 	BASIC_CFLAGS += -DNO_FAST_WORKING_DIRECTORY
 endif

builtin/fast-import.c

@@ -3335,6 +3335,7 @@ static void cat_blob_write(const char *buf, unsigned long size)
 static void cat_blob(struct object_entry *oe, struct object_id *oid)
 {
 	struct strbuf line = STRBUF_INIT;
+	struct iovec iov[3];
 	unsigned long size;
 	enum object_type type = 0;
 	char *buf;
@@ -3368,10 +3369,21 @@ static void cat_blob(struct object_entry *oe, struct object_id *oid)
 	strbuf_reset(&line);
 	strbuf_addf(&line, "%s %s %"PRIuMAX"\n", oid_to_hex(oid),
 		    type_name(type), (uintmax_t)size);
-	cat_blob_write(line.buf, line.len);
+
+	/*
+	 * Write the header, the payload and the trailing newline with a
+	 * single writev(3p) call instead of three separate write(3p) calls.
+	 */
+	iov[0].iov_base = line.buf;
+	iov[0].iov_len = line.len;
+	iov[1].iov_base = buf;
+	iov[1].iov_len = size;
+	iov[2].iov_base = (void *) "\n";
+	iov[2].iov_len = 1;
+
+	if (writev_in_full(cat_blob_fd, iov, ARRAY_SIZE(iov)) < 0)
+		die_errno(_("write to frontend failed"));
 	strbuf_release(&line);
-	cat_blob_write(buf, size);
-	cat_blob_write("\n", 1);
 	if (oe && oe->pack_id == pack_id) {
 		last_blob.offset = oe->idx.offset;
 		strbuf_attach(&last_blob.data, buf, size, size + 1);

compat/posix.h

@@ -148,6 +148,9 @@
 #include <sys/socket.h>
 #include <sys/ioctl.h>
 #include <sys/statvfs.h>
+#ifndef NO_WRITEV
+#include <sys/uio.h>
+#endif
 #include <termios.h>
 #ifndef NO_SYS_SELECT_H
 #include <sys/select.h>
@@ -334,6 +337,17 @@ int git_lstat(const char *, struct stat *);
 ssize_t git_pread(int fd, void *buf, size_t count, off_t offset);
 #endif
 
+#ifdef NO_WRITEV
+#define writev git_writev
+#define iovec git_iovec
+struct git_iovec {
+	void *iov_base;
+	size_t iov_len;
+};
+
+ssize_t git_writev(int fd, const struct iovec *iov, int iovcnt);
+#endif
+
 #ifdef NO_SETENV
 #define setenv gitsetenv
 int gitsetenv(const char *, const char *, int);

compat/writev.c

@@ -0,0 +1,44 @@
+#include "../git-compat-util.h"
+#include "../wrapper.h"
+
+ssize_t git_writev(int fd, const struct iovec *iov, int iovcnt)
+{
+	size_t total_written = 0;
+	size_t sum = 0;
+
+	/*
+	 * According to writev(3p), the syscall shall error with EINVAL in case
+	 * the sum of `iov_len` overflows `ssize_t`.
+	 */
+	for (int i = 0; i < iovcnt; i++) {
+		if (iov[i].iov_len > maximum_signed_value_of_type(ssize_t) ||
+		    iov[i].iov_len + sum > maximum_signed_value_of_type(ssize_t)) {
+			errno = EINVAL;
+			return -1;
+		}
+
+		sum += iov[i].iov_len;
+	}
+
+	for (int i = 0; i < iovcnt; i++) {
+		const char *bytes = iov[i].iov_base;
+		size_t iovec_written = 0;
+
+		while (iovec_written < iov[i].iov_len) {
+			ssize_t bytes_written = xwrite(fd, bytes + iovec_written,
+						       iov[i].iov_len - iovec_written);
+			if (bytes_written < 0) {
+				if (total_written)
+					goto out;
+				return bytes_written;
+			}
+			if (!bytes_written)
+				goto out;
+			iovec_written += bytes_written;
+			total_written += bytes_written;
+		}
+	}
+
+out:
+	return (ssize_t) total_written;
+}

config.mak.uname

@@ -483,6 +483,7 @@ ifeq ($(uname_S),Windows)
 	SANE_TOOL_PATH ?= $(msvc_bin_dir_msys)
 	HAVE_ALLOCA_H = YesPlease
 	NO_PREAD = YesPlease
+	NO_WRITEV = YesPlease
 	NEEDS_CRYPTO_WITH_SSL = YesPlease
 	NO_LIBGEN_H = YesPlease
 	NO_POLL = YesPlease
@@ -697,6 +698,7 @@ ifeq ($(uname_S),MINGW)
 	pathsep = ;
 	HAVE_ALLOCA_H = YesPlease
 	NO_PREAD = YesPlease
+	NO_WRITEV = YesPlease
 	NEEDS_CRYPTO_WITH_SSL = YesPlease
 	NO_LIBGEN_H = YesPlease
 	NO_POLL = YesPlease

contrib/buildsystems/CMakeLists.txt

@@ -378,7 +378,7 @@ endif()
 #function checks
 set(function_checks
 	strcasestr memmem strlcpy strtoimax strtoumax strtoull
-	setenv mkdtemp poll pread memmem)
+	setenv mkdtemp poll pread memmem writev)
 
 #unsetenv,hstrerror are incompatible with windows build
 if(NOT WIN32)
@@ -423,6 +423,10 @@ if(NOT HAVE_MEMMEM)
 	list(APPEND compat_SOURCES compat/memmem.c)
 endif()
 
+if(NOT HAVE_WRITEV)
+	list(APPEND compat_SOURCES compat/writev.c)
+endif()
+
 if(NOT WIN32)
 	if(NOT HAVE_UNSETENV)
 		list(APPEND compat_SOURCES compat/unsetenv.c)

meson.build

@@ -1464,6 +1464,7 @@ checkfuncs = {
   'initgroups' : [],
   'strtoumax' : ['strtoumax.c', 'strtoimax.c'],
   'pread' : ['pread.c'],
+  'writev' : ['writev.c'],
 }
 
 if host_machine.system() == 'windows'

sideband.c

@@ -445,6 +445,7 @@ void send_sideband(int fd, int band, const char *data, ssize_t sz, int packet_ma
 	const char *p = data;
 
 	while (sz) {
+		struct iovec iov[2];
 		unsigned n;
 		char hdr[5];
 
@@ -454,12 +455,19 @@ void send_sideband(int fd, int band, const char *data, ssize_t sz, int packet_ma
 		if (0 <= band) {
 			xsnprintf(hdr, sizeof(hdr), "%04x", n + 5);
 			hdr[4] = band;
-			write_or_die(fd, hdr, 5);
+			iov[0].iov_base = hdr;
+			iov[0].iov_len = 5;
 		} else {
 			xsnprintf(hdr, sizeof(hdr), "%04x", n + 4);
-			write_or_die(fd, hdr, 4);
+			iov[0].iov_base = hdr;
+			iov[0].iov_len = 4;
 		}
-		write_or_die(fd, p, n);
+
+		iov[1].iov_base = (void *) p;
+		iov[1].iov_len = n;
+
+		writev_or_die(fd, iov, ARRAY_SIZE(iov));
+
 		p += n;
 		sz -= n;
 	}

wrapper.c

@@ -323,6 +323,84 @@ ssize_t write_in_full(int fd, const void *buf, size_t count)
 	return total;
 }
 
+ssize_t xwritev(int fd, struct iovec *iov, int iovcnt)
+{
+	size_t allowed = MAX_IO_SIZE;
+	int i;
+
+	/*
+	 * Some platforms define a comparatively small `MAX_IO_SIZE` that
+	 * limits how many bytes can be written with a single call to
+	 * write(3p) or writev(3p); exceeding that limit causes the syscall to
+	 * fail with EINVAL. Just like xwrite() chomps overly large requests
+	 * for write(3p), pretend that the underlying writev(3p) performed a
+	 * short write by only passing along the leading iovec entries that
+	 * fit into that limit.
+	 */
+	for (i = 0; i < iovcnt; i++) {
+		if (iov[i].iov_len > allowed) {
+			/*
+			 * If the first buffer is larger than MAX_IO_SIZE,
+			 * let xwrite() deal with it.
+			 */
+			if (!i)
+				return xwrite(fd, iov->iov_base, iov->iov_len);
+			break;
+		}
+		allowed -= iov[i].iov_len;
+	}
+
+	while (1) {
+		ssize_t bytes_written = writev(fd, iov, i);
+		if (bytes_written < 0) {
+			if (errno == EINTR)
+				continue;
+			if (handle_nonblock(fd, POLLOUT, errno))
+				continue;
+		}
+
+		return bytes_written;
+	}
+}
+
+ssize_t writev_in_full(int fd, struct iovec *iov, int iovcnt)
+{
+	ssize_t total_written = 0;
+
+	while (iovcnt) {
+		ssize_t bytes_written = xwritev(fd, iov, iovcnt);
+		if (bytes_written < 0)
+			return -1;
+		if (!bytes_written) {
+			errno = ENOSPC;
+			return -1;
+		}
+
+		total_written += bytes_written;
+
+		/*
+		 * We first need to discard any iovec entities that have been
+		 * fully written.
+		 */
+		while (iovcnt && (size_t)bytes_written >= iov->iov_len) {
+			bytes_written -= iov->iov_len;
+			iov++;
+			iovcnt--;
+		}
+
+		/*
+		 * Finally, we need to adjust the last iovec in case we have
+		 * performed a partial write.
+		 */
+		if (iovcnt && bytes_written) {
+			iov->iov_base = (char *) iov->iov_base + bytes_written;
+			iov->iov_len -= bytes_written;
+		}
+	}
+
+	return total_written;
+}
+
 ssize_t pread_in_full(int fd, void *buf, size_t count, off_t offset)
 {
 	char *p = buf;

wrapper.h

@@ -16,6 +16,7 @@ void *xmmap_gently(void *start, size_t length, int prot, int flags, int fd, off_
 int xopen(const char *path, int flags, ...);
 ssize_t xread(int fd, void *buf, size_t len);
 ssize_t xwrite(int fd, const void *buf, size_t len);
+ssize_t xwritev(int fd, struct iovec *iov, int iovcnt);
 ssize_t xpread(int fd, void *buf, size_t len, off_t offset);
 int xdup(int fd);
 FILE *xfopen(const char *path, const char *mode);
@@ -47,6 +48,15 @@ ssize_t read_in_full(int fd, void *buf, size_t count);
 ssize_t write_in_full(int fd, const void *buf, size_t count);
 ssize_t pread_in_full(int fd, void *buf, size_t count, off_t offset);
 
+/*
+ * Try to write all iovecs. Returns -1 in case an error occurred with a proper
+ * errno set, the number of bytes written otherwise.
+ *
+ * Note that the iovec will be modified as a result of this call to adjust for
+ * partial writes!
+ */
+ssize_t writev_in_full(int fd, struct iovec *iov, int iovcnt);
+
 static inline ssize_t write_str_in_full(int fd, const char *str)
 {
 	return write_in_full(fd, str, strlen(str));

write-or-die.c

@@ -96,6 +96,14 @@ void write_or_die(int fd, const void *buf, size_t count)
 	}
 }
 
+void writev_or_die(int fd, struct iovec *iov, int iovlen)
+{
+	if (writev_in_full(fd, iov, iovlen) < 0) {
+		check_pipe(errno);
+		die_errno("writev error");
+	}
+}
+
 void fwrite_or_die(FILE *f, const void *buf, size_t count)
 {
 	if (fwrite(buf, 1, count, f) != count)

write-or-die.h

@@ -7,6 +7,7 @@ void fprintf_or_die(FILE *, const char *fmt, ...);
 void fwrite_or_die(FILE *f, const void *buf, size_t count);
 void fflush_or_die(FILE *f);
 void write_or_die(int fd, const void *buf, size_t count);
+void writev_or_die(int fd, struct iovec *iov, int iovlen);
 
 /*
  * These values are used to help identify parts of a repository to fsync.