Merge tag 'trace-v7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-tracegrafted

Pull tracing fixes from Steven Rostedt:

 - Don't destroy user event fields when removal fails

   User event fields are destroyed before the event is removed from
   visibility. But that can fail leaving the still visible event with no
   fields. Move the destroying of the fields to after the event is
   successfully removed from visibility.

 - Initialize function graph state is fork before calling
   copy_exec_state()

   For non-CLONE_VM forks, copy_exec_state() allocates a new
   task_exec_state. If that allocation fails, ftrace_graph_exit_task()
   will free the tasks ret_stack pointer. Since that pointer is still
   using the parent's ret_stack, it mistakenly frees the parent's
   pointer too.

   Call ftrace_graph_init() on the task first which will NULL out the
   new tasks's ret_stack and if the copy fails, it will not free
   anything.

 - Remove FGRAPH_MAX_INDEX

   The macro FGRAPH_MAX_INDEX was added but never used. Remove it.

 - Save ent_size in function graph printing of nested functions

   The function graph tracer needs to look at the next event to see if
   the next event is the return of the current function entry. If it is,
   it prints a single line:

	ktime_get();

   Otherwise it prints it like a nested function:

	tick_nohz_irq_exit() {
	    ktime_get();
	    kcpustat_irq_exit();
	}

   In order to look at the next event, it must save the current event so
   that it has the information to print from it. It saves the event in
   the iterator descriptor called "ent". What it doesn't save is the
   ent_size of the event which is now used to know if the function graph
   arguments are to be printed. The peek doesn't save the size so the
   size used happens to be that of the size of the last event that was
   seen.

   Save the entry event size in the iterator descriptor so that the
   correct size is used.

 - Fix several errors with freeing data in the histogram code

   The histogram code had a lot of leaked or or incorrect accounting
   when failures happen. Correct them.

 - Fix histogram regression of .percent and .graph modifiers

   Up until 6.3 histogram values could have "percent" or "graph"
   modifiers that changed how they were printed. But a change that added
   restricting histograms values from being strings, stack traces and
   other modifiers inadvertently prevented them from using the percent
   and graph modifiers, which were legal use cases for values.

   Put back the percent and graph modifiers.

 - Fix various typos in the comments

 - Set the trace_clock before initializing a histogram with clock
   argument

   The histogram API allows the user to specific which trace clock to
   use via a "clock=" string. The histogram is set up first before the
   clock is checked. If the passed in clock is not valid, it exits
   without fully fixing up the histogram leaving it on the list and a
   use-after-free can trigger.

   Update the clock argument first and if it fails then exit gracefully
   before the histogram trigger is placed on any lists.

 - Restore :mod: trailer after parsing in ftrace_set_clr_event

   The function ftrace_set_clr_event() modifies the parse string and
   needs to put it back to what was passed in. It searches for ":mod:"
   via a strsep() but fails to put back the first ':' in the string.

   Add back the ':' in the passed in string.

 - Take trace_array reference when opening a tracer options file

   The options files are dynamically created and some tracers add their
   own options. When a tracer adds their own list of options, the
   trace_array holding them has an array to hold the list of options for
   each tracer. This array increases in size via a krealloc(), and the
   new entry gets a newly allocated array to hold the options of the new
   tracer being added.

   The element in each entry of the tracer's option array holds a
   pointer back to the trace_array, a pointer to the tracer it is
   associated to, a pointer to the flags of the option.

   The issue is that these arrays are freed when the trace_array is
   freed when its instance it represents is removed from the instances
   directory. There's a race that an open of one of these options files
   can happen when the instance is being removed.

   Add a new helper function to be called by the open function of the
   options file to iterate all existing trace_arrays under a lock and
   find the one that has the given option element in one of it's tracer
   arrays. If found, then update the associated trace_array's reference
   counter to keep it from being freed. If not found, have the open call
   return -ENODEV.

 - Disable interrupts when acquiring the lock in rb_wake_up_waiters()

   The function rb_wake_up_waiters() assumes it will be called in
   interrupt context and does not disable irqs when taking
   cpu_buffer->reader_lock, which can be called in hard interrupt
   context. The issue is in PREEMPT_RT, this function is called in
   thread context leaving this lock open to a deadlock.

   Take the lock with interrupts disabled.

 - Use rcu_assign_pointer() for tmp_ops filter hash

   The tmp_ops used in update_ftrace_direct_mod() assigns its
   filter_hash field directly, but that field is annotated as __rcu and
   sparse complains. Assign it with rcu_assign_pointer()

 - Fix use-after-free in enable_trigger_private_data_free()

   The trace_event_call is accessed through the event_trigger_data's
   trace_event_file pointer to put the trace_event_call on freeing. The
   issue is that the trace_event_file data may have been freed already
   causing a use-after-free. Add a field to the event_trigger_data that
   points directly to the trace_event_call so that it can decrement its
   reference directly without needing to go through the
   trace_event_file.

 - Fix accounting of buffer data remote headers

   trace_buffer_desc_size() and trace_remote_alloc_buffer() undercount
   the number of pages is needed for the asked for size as it doesn't
   take into account the meta data on each page. Add a helper function
   to do the calculation properly and use that in these functions.

 - Catch nr_page_va overflow in ring_buffer_desc sizing

   The number of pages per remote ring buffer is capped by
   ring_buffer_desc::nr_page_va (32 bits). A buffer_size large enough to
   overflow that field would silently allocate a descriptor smaller than
   what was asked for.

 - Do not resize the subbuf order if any per_cpu buffer is disabled

   The mmapping of ring buffers disables resizing the subbuffers, but it
   is done per-cpu whereas the subbuf size change is done for all the
   per_cpu buffers under the buffer->mutex. It could change the size of
   some while the mapping is happening on others. Have the resize of the
   subbuf order check all the per_cpu buffers under the lock to see if
   any of them is disabled before starting and causing an inconsistency
   between buffers that are being mapped.

* tag 'trace-v7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace: (25 commits)
  ring-buffer: Check resize_disabled before publishing the new subbuf order
  tracing/remotes: Catch nr_page_va overflow in ring_buffer_desc sizing
  tracing/remotes: Account for ring buffer page header in size calculation
  tracing: Don't dereference trace_event_file in deferred trigger free
  ftrace: Use rcu_assign_pointer() for tmp_ops filter hash
  ring-buffer: Acquire the lock with irqsave in rb_wake_up_waiters()
  tracing: Take trace_array reference when opening a tracer options file
  tracing: Fix ring_buffer_read_page_size() kernel-doc
  tracing: Restore :mod: trailer after parsing in ftrace_set_clr_event()
  tracing: Fix memory corruption from a "STACKTRACE" histogram key
  tracing: Fix memory corruption from the histogram stacktrace modifier
  tracing: Undo the registration when enabling the histogram trigger fails
  tracing: Take the reference before publishing the named histogram trigger
  tracing: Set the trace clock before registering the histogram trigger
  tracing: Fix typo "preceeded" in comment
  tracing: Fix typo "availabe" in comment
  tracing: Let histogram values keep the percent and graph modifiers
  tracing: Keep the entry count when the histogram stats allocation fails
  tracing: Free histogram the field rejected for a bad modifier
  tracing: Free histogram the var ref when its initialization fails
  ...

1 files changed, 1004 insertions, 0 deletions

diff --git a/tools/include/nolibc/stdio.h b/tools/include/nolibc/stdio.h
new file mode 100644
index 000000000..ebdd413d1
--- /dev/null
+++ b/tools/include/nolibc/stdio.h
@@ -0,0 +1,1004 @@
+/* SPDX-License-Identifier: LGPL-2.1 OR MIT */
+/*
+ * minimal stdio function definitions for NOLIBC
+ * Copyright (C) 2017-2021 Willy Tarreau <w@1wt.eu>
+ */
+
+/* make sure to include all global symbols */
+#include "nolibc.h"
+
+#ifndef _NOLIBC_STDIO_H
+#define _NOLIBC_STDIO_H
+
+#include "std.h"
+#include "arch.h"
+#include "errno.h"
+#include "fcntl.h"
+#include "types.h"
+#include "sys.h"
+#include "stdarg.h"
+#include "stdlib.h"
+#include "string.h"
+#include "compiler.h"
+
+static const char *strerror(int errnum);
+
+#ifndef EOF
+#define EOF (-1)
+#endif
+
+/* Buffering mode used by setvbuf.  */
+#define _IOFBF 0	/* Fully buffered. */
+#define _IOLBF 1	/* Line buffered. */
+#define _IONBF 2	/* No buffering. */
+
+/* just define FILE as a non-empty type. The value of the pointer gives
+ * the FD: FILE=~fd for fd>=0 or NULL for fd<0. This way positive FILE
+ * are immediately identified as abnormal entries (i.e. possible copies
+ * of valid pointers to something else).
+ */
+typedef struct FILE {
+	char dummy[1];
+} FILE;
+
+static __attribute__((unused)) FILE* const stdin  = (FILE*)(intptr_t)~STDIN_FILENO;
+static __attribute__((unused)) FILE* const stdout = (FILE*)(intptr_t)~STDOUT_FILENO;
+static __attribute__((unused)) FILE* const stderr = (FILE*)(intptr_t)~STDERR_FILENO;
+
+/* provides a FILE* equivalent of fd. The mode is ignored. */
+static __attribute__((unused))
+FILE *fdopen(int fd, const char *mode __attribute__((unused)))
+{
+	if (fd < 0) {
+		SET_ERRNO(EBADF);
+		return NULL;
+	}
+	return (FILE*)(intptr_t)~fd;
+}
+
+static __attribute__((unused))
+FILE *fopen(const char *pathname, const char *mode)
+{
+	int flags, fd;
+
+	switch (*mode) {
+	case 'r':
+		flags = O_RDONLY;
+		break;
+	case 'w':
+		flags = O_WRONLY | O_CREAT | O_TRUNC;
+		break;
+	case 'a':
+		flags = O_WRONLY | O_CREAT | O_APPEND;
+		break;
+	default:
+		SET_ERRNO(EINVAL); return NULL;
+	}
+
+	if (mode[1] == '+')
+		flags = (flags & ~(O_RDONLY | O_WRONLY)) | O_RDWR;
+
+	fd = open(pathname, flags, 0666);
+	return fdopen(fd, mode);
+}
+
+/* provides the fd of stream. */
+static __attribute__((unused))
+int fileno(FILE *stream)
+{
+	intptr_t i = (intptr_t)stream;
+
+	if (i >= 0) {
+		SET_ERRNO(EBADF);
+		return -1;
+	}
+	return ~i;
+}
+
+/* flush a stream. */
+static __attribute__((unused))
+int fflush(FILE *stream)
+{
+	intptr_t i = (intptr_t)stream;
+
+	/* NULL is valid here. */
+	if (i > 0) {
+		SET_ERRNO(EBADF);
+		return -1;
+	}
+
+	/* Don't do anything, nolibc does not support buffering. */
+	return 0;
+}
+
+/* flush a stream. */
+static __attribute__((unused))
+int fclose(FILE *stream)
+{
+	intptr_t i = (intptr_t)stream;
+
+	if (i >= 0) {
+		SET_ERRNO(EBADF);
+		return -1;
+	}
+
+	if (close(~i))
+		return EOF;
+
+	return 0;
+}
+
+/* getc(), fgetc(), getchar() */
+
+#define getc(stream) fgetc(stream)
+
+static __attribute__((unused))
+int fgetc(FILE* stream)
+{
+	unsigned char ch;
+
+	if (read(fileno(stream), &ch, 1) <= 0)
+		return EOF;
+	return ch;
+}
+
+static __attribute__((unused))
+int getchar(void)
+{
+	return fgetc(stdin);
+}
+
+
+/* putc(), fputc(), putchar() */
+
+#define putc(c, stream) fputc(c, stream)
+
+static __attribute__((unused))
+int fputc(int c, FILE* stream)
+{
+	unsigned char ch = c;
+
+	if (write(fileno(stream), &ch, 1) <= 0)
+		return EOF;
+	return ch;
+}
+
+static __attribute__((unused))
+int putchar(int c)
+{
+	return fputc(c, stdout);
+}
+
+
+/* fwrite(), fread(), puts(), fputs(). Note that puts() emits '\n' but not fputs(). */
+
+/* internal fwrite()-like function which only takes a size and returns 0 on
+ * success or EOF on error. It automatically retries on short writes.
+ */
+static __attribute__((unused))
+int _fwrite(const void *buf, size_t size, FILE *stream)
+{
+	ssize_t ret;
+	int fd = fileno(stream);
+
+	while (size) {
+		ret = write(fd, buf, size);
+		if (ret <= 0)
+			return EOF;
+		size -= ret;
+		buf += ret;
+	}
+	return 0;
+}
+
+static __attribute__((unused))
+size_t fwrite(const void *s, size_t size, size_t nmemb, FILE *stream)
+{
+	size_t written;
+
+	for (written = 0; written < nmemb; written++) {
+		if (_fwrite(s, size, stream) != 0)
+			break;
+		s += size;
+	}
+	return written;
+}
+
+/* internal fread()-like function which only takes a size and returns 0 on
+ * success or EOF on error. It automatically retries on short reads.
+ */
+static __attribute__((unused))
+int _fread(void *buf, size_t size, FILE *stream)
+{
+	int fd = fileno(stream);
+	ssize_t ret;
+
+	while (size) {
+		ret = read(fd, buf, size);
+		if (ret <= 0)
+			return EOF;
+		size -= ret;
+		buf += ret;
+	}
+	return 0;
+}
+
+static __attribute__((unused))
+size_t fread(void *s, size_t size, size_t nmemb, FILE *stream)
+{
+	size_t nread;
+
+	for (nread = 0; nread < nmemb; nread++) {
+		if (_fread(s, size, stream) != 0)
+			break;
+		s += size;
+	}
+	return nread;
+}
+
+static __attribute__((unused))
+int fputs(const char *s, FILE *stream)
+{
+	return _fwrite(s, strlen(s), stream);
+}
+
+static __attribute__((unused))
+int puts(const char *s)
+{
+	if (fputs(s, stdout) == EOF)
+		return EOF;
+	return putchar('\n');
+}
+
+
+/* fgets() */
+static __attribute__((unused))
+char *fgets(char *s, int size, FILE *stream)
+{
+	int ofs;
+	int c;
+
+	for (ofs = 0; ofs + 1 < size;) {
+		c = fgetc(stream);
+		if (c == EOF)
+			break;
+		s[ofs++] = c;
+		if (c == '\n')
+			break;
+	}
+	if (ofs < size)
+		s[ofs] = 0;
+	return ofs ? s : NULL;
+}
+
+
+/* fseek */
+static __attribute__((unused))
+int fseek(FILE *stream, long offset, int whence)
+{
+	int fd = fileno(stream);
+	off_t ret;
+
+	ret = lseek(fd, offset, whence);
+
+	/* lseek() and fseek() differ in that lseek returns the new
+	 * position or -1, fseek() returns either 0 or -1.
+	 */
+	if (ret >= 0)
+		return 0;
+
+	return -1;
+}
+
+
+/* printf(). Supports most of the normal integer and string formats.
+ *  - %[#0-+ ][width|*[.precision|*}][{l,t,z,ll,L,j,q}]{c,d,i,u,o,x,X,p,s,m,%}
+ *  - %% generates a single %
+ *  - %m outputs strerror(errno).
+ *  - %X outputs a..f the same as %x.
+ *  - No support for floating point or wide characters.
+ *  - Invalid formats are copied to the output buffer.
+ *
+ * Called by vfprintf() and snprintf() to do the actual formatting.
+ * The callers provide a callback function to save the formatted data.
+ * The callback function is called multiple times:
+ *  - for each group of literal characters in the format string.
+ *  - for field padding.
+ *  - for each conversion specifier.
+ *  - with (NULL, 0) at the end of the __nolibc_printf.
+ * If the callback returns non-zero __nolibc_printf() immediately returns -1.
+ */
+
+typedef int (*__nolibc_printf_cb)(void *state, const char *buf, size_t size);
+
+/* This code uses 'flag' variables that are indexed by the low 6 bits
+ * of characters to optimise checks for multiple characters.
+ *
+ * _NOLIBC_PF_FLAGS_CONTAIN(flags, 'a', 'b'. ...)
+ * returns non-zero if the bit for any of the specified characters is set.
+ *
+ * _NOLIBC_PF_CHAR_IS_ONE_OF(ch, 'a', 'b'. ...)
+ * returns the flag bit for ch if it is one of the specified characters.
+ * All the characters must be in the same 32 character block (non-alphabetic,
+ * upper case, or lower case) of the ASCII character set.
+ */
+#define _NOLIBC_PF_FLAG(ch) (1u << ((ch) & 0x1f))
+#define _NOLIBC_PF_FLAG_NZ(ch) ((ch) ? _NOLIBC_PF_FLAG(ch) : 0)
+#define _NOLIBC_PF_FLAG8(cmp_1, cmp_2, cmp_3, cmp_4, cmp_5, cmp_6, cmp_7, cmp_8, ...) \
+	(_NOLIBC_PF_FLAG_NZ(cmp_1) | _NOLIBC_PF_FLAG_NZ(cmp_2) | \
+	 _NOLIBC_PF_FLAG_NZ(cmp_3) | _NOLIBC_PF_FLAG_NZ(cmp_4) | \
+	 _NOLIBC_PF_FLAG_NZ(cmp_5) | _NOLIBC_PF_FLAG_NZ(cmp_6) | \
+	 _NOLIBC_PF_FLAG_NZ(cmp_7) | _NOLIBC_PF_FLAG_NZ(cmp_8))
+#define _NOLIBC_PF_FLAGS_CONTAIN(flags, ...) \
+	((flags) & _NOLIBC_PF_FLAG8(__VA_ARGS__, 0, 0, 0, 0, 0, 0, 0))
+#define _NOLIBC_PF_CHAR_IS_ONE_OF(ch, cmp_1, ...) \
+	((unsigned int)(ch) - (cmp_1 & 0xe0) > 0x1f ? 0 : \
+		_NOLIBC_PF_FLAGS_CONTAIN(_NOLIBC_PF_FLAG(ch), cmp_1, __VA_ARGS__))
+
+static __attribute__((unused, format(printf, 3, 0)))
+int __nolibc_printf(__nolibc_printf_cb cb, void *state, const char *fmt, va_list args)
+{
+	char ch;
+	unsigned long long v;
+	long long signed_v;
+	int written, width, precision, len;
+	unsigned int flags, ch_flag;
+	char outbuf[2 + 31 + 22 + 1];
+	char *out;
+	const char *outstr;
+	unsigned int sign_prefix;
+	int got_width;
+
+	written = 0;
+	while (1) {
+		outstr = fmt;
+		ch = *fmt++;
+		if (!ch)
+			break;
+
+		width = 0;
+		flags = 0;
+		if (ch != '%') {
+			while (*fmt && *fmt != '%')
+				fmt++;
+			/* Output characters from the format string. */
+			len = fmt - outstr;
+			goto do_output;
+		}
+
+		/* we're in a format sequence */
+
+		/* Conversion flag characters */
+		while (1) {
+			ch = *fmt++;
+			ch_flag = _NOLIBC_PF_CHAR_IS_ONE_OF(ch, ' ', '#', '+', '-', '0');
+			if (!ch_flag)
+				break;
+			flags |= ch_flag;
+		}
+
+		/* Width and precision */
+		for (got_width = 0;; ch = *fmt++) {
+			if (ch == '*') {
+				precision = va_arg(args, int);
+				ch = *fmt++;
+			} else {
+				for (precision = 0; ch >= '0' && ch <= '9'; ch = *fmt++)
+					precision = precision * 10 + (ch - '0');
+			}
+			if (got_width)
+				break;
+			width = precision;
+			if (ch != '.') {
+				/* Default precision for strings */
+				precision = -1;
+				break;
+			}
+			got_width = 1;
+		}
+		/* A negative width (e.g. from "%*s") requests left justify. */
+		if (width < 0) {
+			width = -width;
+			flags |= _NOLIBC_PF_FLAG('-');
+		}
+
+		/* Length modifier.
+		 * They miss the conversion flags characters " #+-0" so can go into flags.
+		 * Change both L and ll to j (all always 64bit).
+		 */
+		if (ch == 'L')
+			ch = 'j';
+		ch_flag = _NOLIBC_PF_CHAR_IS_ONE_OF(ch, 'l', 't', 'z', 'j', 'q');
+		if (ch_flag != 0) {
+			if (ch == 'l' && fmt[0] == 'l') {
+				fmt++;
+				ch_flag = _NOLIBC_PF_FLAG('j');
+			}
+			flags |= ch_flag;
+			ch = *fmt++;
+		}
+
+		/* Conversion specifiers. */
+
+		/* Numeric and pointer conversion specifiers.
+		 *
+		 * Use an explicit bound check (rather than _NOLIBC_PF_CHAR_IS_ONE_OF())
+		 * so that 'X' can be allowed through.
+		 * 'X' gets treated and 'x' because _NOLIBC_PF_FLAG() returns the same
+		 * value for both.
+		 *
+		 * We need to check for "%p" or "%#x" later, merging here gives better code.
+		 * But '#' collides with 'c' so shift right.
+		 */
+		ch_flag = _NOLIBC_PF_FLAG(ch) | (flags & _NOLIBC_PF_FLAG('#')) >> 1;
+		if (((ch >= 'a' && ch <= 'z') || ch == 'X') &&
+		    _NOLIBC_PF_FLAGS_CONTAIN(ch_flag, 'c', 'd', 'i', 'u', 'o', 'x', 'p', 's')) {
+			/* 'long' is needed for pointer/string conversions and ltz lengths.
+			 * A single test can be used provided 'p' (the same bit as '0')
+			 * is masked from flags.
+			 */
+			if (_NOLIBC_PF_FLAGS_CONTAIN(ch_flag | (flags & ~_NOLIBC_PF_FLAG('p')),
+						     'p', 's', 'l', 't', 'z')) {
+				v = va_arg(args, unsigned long);
+				signed_v = (long)v;
+			} else if (_NOLIBC_PF_FLAGS_CONTAIN(flags, 'j', 'q')) {
+				v = va_arg(args, unsigned long long);
+				signed_v = v;
+			} else {
+				v = va_arg(args, unsigned int);
+				signed_v = (int)v;
+			}
+
+			if (ch == 'c') {
+				/* "%c" - single character. */
+				outbuf[0] = v;
+				len = 1;
+				outstr = outbuf;
+				goto do_output;
+			}
+
+			if (ch == 's') {
+				/* "%s" - character string. */
+				outstr = (const char  *)(uintptr_t)v;
+				if (!outstr) {
+					outstr = "(null)";
+					/* Match glibc, nothing output if precision too small */
+					len = precision < 0 || precision >= 6 ? 6 : 0;
+					goto do_output;
+				}
+				goto do_strlen_output;
+			}
+
+			/* The 'sign_prefix' can be zero, one or two ("0x") characters.
+			 * Prepended least significant byte first stopping on a zero byte.
+			 */
+			sign_prefix = 0;
+
+			if (_NOLIBC_PF_FLAGS_CONTAIN(ch_flag, 'd', 'i')) {
+				/* "%d" and "%i" - signed decimal numbers. */
+				if (signed_v < 0) {
+					sign_prefix = '-';
+					v = -(signed_v + 1);
+					v++;
+				} else if (_NOLIBC_PF_FLAGS_CONTAIN(flags, '+')) {
+					sign_prefix = '+';
+				} else if (_NOLIBC_PF_FLAGS_CONTAIN(flags, ' ')) {
+					sign_prefix = ' ';
+				}
+			} else {
+				/* "#o" requires that the output always starts with a '0'.
+				 * This needs another check after any zero padding to avoid
+				 * adding an extra leading '0'.
+				 */
+				if (_NOLIBC_PF_FLAGS_CONTAIN(ch_flag, 'o') &&
+						_NOLIBC_PF_FLAGS_CONTAIN(ch_flag, '#' - 1))
+					sign_prefix = '0';
+			}
+
+			/* The value is converted offset into the buffer so that
+			 * 31 zero pad characters and the sign/prefix can be added in front.
+			 * The longest digit string is 22 + 1 for octal conversions.
+			 */
+			out = outbuf + 2 + 31;
+
+			if (v == 0) {
+				/* There are special rules for zero. */
+				if (_NOLIBC_PF_FLAGS_CONTAIN(ch_flag, 'p')) {
+					/* "%p" match glibc, precision is ignored */
+					outstr = "(nil)";
+					len = 5;
+					goto do_output;
+				}
+				if (!precision) {
+					/* Explicit %nn.0d, no digits output (except for %#.0o) */
+					len = 0;
+					goto prepend_sign;
+				}
+				/* All other formats (including "%#x") just output "0". */
+				out[0] = '0';
+				len = 1;
+			} else {
+				/* Convert the number to ascii in the required base. */
+				unsigned long long recip;
+				unsigned int base;
+				if (_NOLIBC_PF_FLAGS_CONTAIN(ch_flag, 'd', 'i', 'u')) {
+					base = 10;
+					recip = _NOLIBC_U64TOA_RECIP(10);
+				} else if (_NOLIBC_PF_FLAGS_CONTAIN(ch_flag, 'o')) {
+					base = 8;
+					recip = _NOLIBC_U64TOA_RECIP(8);
+				} else {
+					base = 16;
+					recip = _NOLIBC_U64TOA_RECIP(16);
+					if (_NOLIBC_PF_FLAGS_CONTAIN(ch_flag, 'p', '#' - 1)) {
+						/* "%p" and "%#x" need "0x" prepending. */
+						sign_prefix = '0' << 8 | 'x';
+					}
+				}
+				len = _nolibc_u64toa_base(v, out, base, recip);
+			}
+
+			/* Add zero padding */
+			if (precision < 0) {
+				/* No explicit precision (or negative from "%.*s"). */
+				if (!_NOLIBC_PF_FLAGS_CONTAIN(flags, '0'))
+					goto no_zero_padding;
+				if (_NOLIBC_PF_FLAGS_CONTAIN(flags, '-'))
+					/* Left justify overrides zero pad */
+					goto no_zero_padding;
+				/* eg "%05d", Zero pad to field width less sign.
+				 * Note that precision can end up negative so all
+				 * the variables have to be 'signed int'.
+				 */
+				precision = width;
+				if (sign_prefix) {
+					precision--;
+					if (sign_prefix >= 256)
+						precision--;
+				}
+			}
+			if (precision > 31)
+				/* Don't run off the start of outbuf[], arbitrary limit
+				 * longer than the longest number field. */
+				precision = 31;
+			for (; len < precision; len++) {
+				/* Stop gcc generating horrid code and memset(). */
+				_NOLIBC_OPTIMIZER_HIDE_VAR(len);
+				*--out = '0';
+			}
+no_zero_padding:
+
+			/* %#o has set sign_prefix to '0', but we don't want so add an extra
+			 * leading zero here.
+			 * Since the only other byte values of sign_prefix are ' ', '+' and '-'
+			 * it is enough to check that out[] doesn't already start with sign_prefix.
+			 */
+			if (sign_prefix - *out) {
+prepend_sign:
+				/* Add the 0, 1 or 2 ("0x") sign/prefix characters at the front. */
+				for (; sign_prefix; sign_prefix >>= 8) {
+					/* Force gcc to increment len inside the loop. */
+					_NOLIBC_OPTIMIZER_HIDE_VAR(len);
+					len++;
+					*--out = sign_prefix;
+				}
+			}
+			outstr = out;
+			goto do_output;
+		}
+
+		if (ch == 'm') {
+#ifdef NOLIBC_IGNORE_ERRNO
+			outstr = "unknown error";
+#else
+			outstr = strerror(errno);
+#endif /* NOLIBC_IGNORE_ERRNO */
+			goto do_strlen_output;
+		}
+
+		if (ch != '%') {
+			/* Invalid format: back up to output the format characters */
+			fmt = outstr + 1;
+			/* and output a '%' now. */
+		}
+		/* %% is documented as a 'conversion specifier'.
+		 * Any flags, precision or length modifier are ignored.
+		 */
+		len = 1;
+		width = 0;
+		outstr = fmt - 1;
+		goto do_output;
+
+do_strlen_output:
+		/* Open coded strnlen() (slightly smaller). */
+		for (len = 0; precision < 0 || len < precision; len++)
+			if (!outstr[len])
+				break;
+
+do_output:
+		written += len;
+
+		/* Stop gcc back-merging this code into one of the conditionals above. */
+		_NOLIBC_OPTIMIZER_HIDE_VAR(len);
+
+		/* Output the characters on the required side of any padding. */
+		width -= len;
+		flags = _NOLIBC_PF_FLAGS_CONTAIN(flags, '-');
+		if (flags && cb(state, outstr, len) != 0)
+			return -1;
+		while (width > 0) {
+			/* Output pad in 16 byte blocks with the small block first. */
+			int pad_len = ((width - 1) & 15) + 1;
+			width -= pad_len;
+			written += pad_len;
+			if (cb(state, "                ", pad_len) != 0)
+				return -1;
+		}
+		if (!flags && cb(state, outstr, len) != 0)
+			return -1;
+	}
+
+	/* Request a final '\0' be added to the snprintf() output.
+	 * This may be the only call of the cb() function.
+	 */
+	if (cb(state, NULL, 0) != 0)
+		return -1;
+
+	return written;
+}
+
+static int __nolibc_fprintf_cb(void *stream, const char *buf, size_t size)
+{
+	return _fwrite(buf, size, stream);
+}
+
+static __attribute__((unused, format(printf, 2, 0)))
+int vfprintf(FILE *stream, const char *fmt, va_list args)
+{
+	return __nolibc_printf(__nolibc_fprintf_cb, stream, fmt, args);
+}
+
+static __attribute__((unused, format(printf, 1, 0)))
+int vprintf(const char *fmt, va_list args)
+{
+	return vfprintf(stdout, fmt, args);
+}
+
+static __attribute__((unused, format(printf, 2, 3)))
+int fprintf(FILE *stream, const char *fmt, ...)
+{
+	va_list args;
+	int ret;
+
+	va_start(args, fmt);
+	ret = vfprintf(stream, fmt, args);
+	va_end(args);
+	return ret;
+}
+
+static __attribute__((unused, format(printf, 1, 2)))
+int printf(const char *fmt, ...)
+{
+	va_list args;
+	int ret;
+
+	va_start(args, fmt);
+	ret = vfprintf(stdout, fmt, args);
+	va_end(args);
+	return ret;
+}
+
+static __attribute__((unused, format(printf, 2, 0)))
+int vdprintf(int fd, const char *fmt, va_list args)
+{
+	FILE *stream;
+
+	stream = fdopen(fd, NULL);
+	if (!stream)
+		return -1;
+	/* Technically 'stream' is leaked, but as it's only a wrapper around 'fd' that is fine */
+	return vfprintf(stream, fmt, args);
+}
+
+static __attribute__((unused, format(printf, 2, 3)))
+int dprintf(int fd, const char *fmt, ...)
+{
+	va_list args;
+	int ret;
+
+	va_start(args, fmt);
+	ret = vdprintf(fd, fmt, args);
+	va_end(args);
+
+	return ret;
+}
+
+struct __nolibc_sprintf_cb_state {
+	char *buf;
+	size_t space;
+};
+
+static int __nolibc_sprintf_cb(void *v_state, const char *buf, size_t size)
+{
+	struct __nolibc_sprintf_cb_state *state = v_state;
+	size_t space = state->space;
+	char *tgt;
+
+	/* Truncate the request to fit in the output buffer space.
+	 * The last byte is reserved for the terminating '\0'.
+	 * state->space can only be zero for snprintf(NULL, 0, fmt, args)
+	 * so this normally lets through calls with 'size == 0'.
+	 */
+	if (size >= space) {
+		if (space <= 1)
+			return 0;
+		size = space - 1;
+	}
+	tgt = state->buf;
+
+	/* __nolibc_printf() ends with cb(state, NULL, 0) to request the output
+	 * buffer be '\0' terminated.
+	 * That will be the only cb() call for, eg, snprintf(buf, sz, "").
+	 * Zero lengths can occur at other times (eg "%s" for an empty string).
+	 * Unconditionally write the '\0' byte to reduce code size, it is
+	 * normally overwritten by the data being output.
+	 * There is no point adding a '\0' after copied data - there is always
+	 * another call.
+	 */
+	*tgt = '\0';
+	if (size) {
+		state->space = space - size;
+		state->buf = tgt + size;
+		memcpy(tgt, buf, size);
+	}
+
+	return 0;
+}
+
+static __attribute__((unused, format(printf, 3, 0)))
+int vsnprintf(char *buf, size_t size, const char *fmt, va_list args)
+{
+	struct __nolibc_sprintf_cb_state state = { .buf = buf, .space = size };
+
+	return __nolibc_printf(__nolibc_sprintf_cb, &state, fmt, args);
+}
+
+static __attribute__((unused, format(printf, 3, 4)))
+int snprintf(char *buf, size_t size, const char *fmt, ...)
+{
+	va_list args;
+	int ret;
+
+	va_start(args, fmt);
+	ret = vsnprintf(buf, size, fmt, args);
+	va_end(args);
+
+	return ret;
+}
+
+static __attribute__((unused, format(printf, 2, 0)))
+int vsprintf(char *buf, const char *fmt, va_list args)
+{
+	return vsnprintf(buf, SIZE_MAX, fmt, args);
+}
+
+static __attribute__((unused, format(printf, 2, 3)))
+int sprintf(char *buf, const char *fmt, ...)
+{
+	va_list args;
+	int ret;
+
+	va_start(args, fmt);
+	ret = vsprintf(buf, fmt, args);
+	va_end(args);
+
+	return ret;
+}
+
+static __attribute__((unused, format(printf, 2, 0)))
+int __nolibc_vasprintf(char **strp, const char *fmt, va_list args1, va_list args2)
+{
+	int len1, len2;
+	char *buf;
+
+	len1 = vsnprintf(NULL, 0, fmt, args1);
+	if (len1 < 0)
+		return -1;
+
+	buf = malloc(len1 + 1);
+	if (!buf)
+		return -1;
+
+	len2 = vsnprintf(buf, len1 + 1, fmt, args2);
+	if (len2 < 0) {
+		free(buf);
+		return -1;
+	}
+
+	*strp = buf;
+	return len1;
+}
+
+static __attribute__((unused, format(printf, 2, 0)))
+int vasprintf(char **strp, const char *fmt, va_list args)
+{
+	va_list args2;
+	int ret;
+
+	va_copy(args2, args);
+	ret = __nolibc_vasprintf(strp, fmt, args, args2);
+	va_end(args2);
+
+	return ret;
+}
+
+static __attribute__((unused, format(printf, 2, 3)))
+int asprintf(char **strp, const char *fmt, ...)
+{
+	va_list args;
+	int ret;
+
+	va_start(args, fmt);
+	ret = vasprintf(strp, fmt, args);
+	va_end(args);
+
+	return ret;
+}
+
+static __attribute__((unused))
+int vsscanf(const char *str, const char *format, va_list args)
+{
+	uintmax_t uval;
+	intmax_t ival;
+	int base;
+	char *endptr;
+	int matches;
+	int lpref;
+
+	matches = 0;
+
+	while (1) {
+		if (*format == '%') {
+			/* start of pattern */
+			lpref = 0;
+			format++;
+
+			if (*format == 'l') {
+				/* same as in printf() */
+				lpref = 1;
+				format++;
+				if (*format == 'l') {
+					lpref = 2;
+					format++;
+				}
+			}
+
+			if (*format == '%') {
+				/* literal % */
+				if ('%' != *str)
+					goto done;
+				str++;
+				format++;
+				continue;
+			} else if (*format == 'd') {
+				ival = strtoll(str, &endptr, 10);
+				if (lpref == 0)
+					*va_arg(args, int *) = ival;
+				else if (lpref == 1)
+					*va_arg(args, long *) = ival;
+				else if (lpref == 2)
+					*va_arg(args, long long *) = ival;
+			} else if (*format == 'u' || *format == 'x' || *format == 'X') {
+				base = *format == 'u' ? 10 : 16;
+				uval = strtoull(str, &endptr, base);
+				if (lpref == 0)
+					*va_arg(args, unsigned int *) = uval;
+				else if (lpref == 1)
+					*va_arg(args, unsigned long *) = uval;
+				else if (lpref == 2)
+					*va_arg(args, unsigned long long *) = uval;
+			} else if (*format == 'p') {
+				*va_arg(args, void **) = (void *)strtoul(str, &endptr, 16);
+			} else {
+				SET_ERRNO(EILSEQ);
+				goto done;
+			}
+
+			format++;
+			str = endptr;
+			matches++;
+
+		} else if (*format == '\0') {
+			goto done;
+		} else if (isspace(*format)) {
+			/* skip spaces in format and str */
+			while (isspace(*format))
+				format++;
+			while (isspace(*str))
+				str++;
+		} else if (*format == *str) {
+			/* literal match */
+			format++;
+			str++;
+		} else {
+			if (!matches)
+				matches = EOF;
+			goto done;
+		}
+	}
+
+done:
+	return matches;
+}
+
+static __attribute__((unused, format(scanf, 2, 3)))
+int sscanf(const char *str, const char *format, ...)
+{
+	va_list args;
+	int ret;
+
+	va_start(args, format);
+	ret = vsscanf(str, format, args);
+	va_end(args);
+	return ret;
+}
+
+static __attribute__((unused))
+void perror(const char *msg)
+{
+#ifdef NOLIBC_IGNORE_ERRNO
+	fprintf(stderr, "%s%sunknown error\n", (msg && *msg) ? msg : "", (msg && *msg) ? ": " : "");
+#else
+	fprintf(stderr, "%s%serrno=%d\n", (msg && *msg) ? msg : "", (msg && *msg) ? ": " : "", errno);
+#endif
+}
+
+static __attribute__((unused))
+int setvbuf(FILE *stream __attribute__((unused)),
+	    char *buf __attribute__((unused)),
+	    int mode,
+	    size_t size __attribute__((unused)))
+{
+	/*
+	 * nolibc does not support buffering so this is a nop. Just check mode
+	 * is valid as required by the spec.
+	 */
+	switch (mode) {
+	case _IOFBF:
+	case _IOLBF:
+	case _IONBF:
+		break;
+	default:
+		return EOF;
+	}
+
+	return 0;
+}
+
+static __attribute__((unused))
+int strerror_r(int errnum, char *buf, size_t buflen)
+{
+	if (buflen < 18)
+		return ERANGE;
+
+	__builtin_memcpy(buf, "errno=", 6);
+	i64toa_r(errnum, buf + 6);
+	return 0;
+}
+
+static __attribute__((unused))
+const char *strerror(int errnum)
+{
+	static char buf[18];
+	char *b = buf;
+
+	/* Force gcc to use 'register offset' to access buf[]. */
+	_NOLIBC_OPTIMIZER_HIDE_VAR(b);
+
+	/* Use strerror_r() to avoid having the only .data in small programs. */
+	strerror_r(errnum, b, sizeof(buf));
+
+	return b;
+}
+
+#endif /* _NOLIBC_STDIO_H */