kernel: Add string view and string builder

This makes write/read calls that use strings to communicate much
simpler and less error prone.

11 files changed, 482 insertions, 44 deletions

diff --git a/kernel/Makefile b/kernel/Makefile
index a27b108..34b9b30 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -1,6 +1,6 @@
 CC="i686-sb-gcc"
 AS="i686-sb-as"
-OBJ = arch/i386/boot.o init/kernel.o cpu/gdt.o cpu/reload_gdt.o cpu/idt.o cpu/io.o libc/stdio/print.o drivers/keyboard.o log.o drivers/pit.o libc/string/memcpy.o libc/string/strlen.o libc/string/memcmp.o drivers/ata.o libc/string/memset.o cpu/syscall.o  read_eip.o libc/exit/assert.o process.o libc/string/strcpy.o arch/i386/mmu.o kmalloc.o fs/ext2.o fs/vfs.o fs/devfs.o cpu/spinlock.o random.o libc/string/strcmp.o crypto/ChaCha20/chacha20.o crypto/SHA1/sha1.o fs/tmpfs.o libc/string/isequal.o drivers/pst.o kubsan.o drivers/serial.o socket.o poll.o fs/fifo.o hashmap/hashmap.o fs/shm.o elf.o sched/scheduler.o libc/string/copy.o drivers/mouse.o libc/string/strlcpy.o libc/string/strcat.o drivers/vbe.o drivers/pci.o drivers/rtl8139.o network/ethernet.o network/arp.o network/bytes.o network/ipv4.o network/udp.o signal.o network/tcp.o drivers/ahci.o crypto/xoshiro256plusplus/xoshiro256plusplus.o arch/i386/interrupts.o cpu/isr.o lib/stack.o lib/buffered_write.o lib/list.o cpu/arch_inst.o cpu/int_syscall.o lib/ringbuffer.o lib/relist.o arch/i386/tsc.o arch/i386/asm_tsc.o drivers/cmos.o timer.o queue.o fonts.o drivers/ac97.o audio.o libc/ctype/isdigit.o libc/stdlib/parsenum.o libc/ctype/tolower.o
+OBJ = arch/i386/boot.o init/kernel.o cpu/gdt.o cpu/reload_gdt.o cpu/idt.o cpu/io.o libc/stdio/print.o drivers/keyboard.o log.o drivers/pit.o libc/string/memcpy.o libc/string/strlen.o libc/string/memcmp.o drivers/ata.o libc/string/memset.o cpu/syscall.o  read_eip.o libc/exit/assert.o process.o libc/string/strcpy.o arch/i386/mmu.o kmalloc.o fs/ext2.o fs/vfs.o fs/devfs.o cpu/spinlock.o random.o libc/string/strcmp.o crypto/ChaCha20/chacha20.o crypto/SHA1/sha1.o fs/tmpfs.o libc/string/isequal.o drivers/pst.o kubsan.o drivers/serial.o socket.o poll.o fs/fifo.o hashmap/hashmap.o fs/shm.o elf.o sched/scheduler.o libc/string/copy.o drivers/mouse.o libc/string/strlcpy.o libc/string/strcat.o drivers/vbe.o drivers/pci.o drivers/rtl8139.o network/ethernet.o network/arp.o network/bytes.o network/ipv4.o network/udp.o signal.o network/tcp.o drivers/ahci.o crypto/xoshiro256plusplus/xoshiro256plusplus.o arch/i386/interrupts.o cpu/isr.o lib/stack.o lib/buffered_write.o lib/list.o cpu/arch_inst.o cpu/int_syscall.o lib/ringbuffer.o lib/relist.o arch/i386/tsc.o arch/i386/asm_tsc.o drivers/cmos.o timer.o queue.o fonts.o drivers/ac97.o audio.o libc/ctype/isdigit.o libc/stdlib/parsenum.o libc/ctype/tolower.o lib/sv.o lib/sb.o libc/string/memmove.o libc/ctype/isspace.o
 CFLAGS = -std=c99 -O0 -fsanitize=vla-bound,shift-exponent,pointer-overflow,shift,signed-integer-overflow,bounds -ggdb -ffreestanding -Wall -Wextra -Wno-int-conversion -Wno-unused-parameter -Werror -mgeneral-regs-only -Wimplicit-fallthrough -I./libc/include/ -I. -Wno-pointer-sign -DKERNEL
 LDFLAGS=
 #CFLAGS = -std=c99 -O3 -flto -ggdb -ffreestanding -Wall -Wextra -Wno-int-conversion -Wno-unused-parameter -Werror -mgeneral-regs-only -Wimplicit-fallthrough -I./libc/include/ -I. -Wno-pointer-sign -DKERNEL
diff --git a/kernel/audio.c b/kernel/audio.c
index d8fcd3c..fe78938 100644
--- a/kernel/audio.c
+++ b/kernel/audio.c
@@ -5,6 +5,7 @@
 #include <drivers/ac97.h>
 #include <errno.h>
 #include <fs/devfs.h>
+#include <lib/sv.h>
 #include <math.h>
 
 int audio_write(u8 *buffer, u64 offset, u64 len, vfs_fd_t *fd) {
@@ -25,24 +26,18 @@ int audio_can_write(vfs_inode_t *inode) {
 int volume_write(u8 *buffer, u64 offset, u64 len, vfs_fd_t *fd) {
   (void)offset;
   (void)fd;
-  int volume = 0;
-
-  size_t i = 0;
-  for (; i < len; i++) {
-    u8 c = buffer[i];
-    if (!isdigit(c)) {
-      break;
-    }
-    volume *= 10;
-    volume += c - '0';
-    if (volume > 100) {
-      volume = 100;
-      break;
-    }
-  }
+  struct sv string_view = sv_init(buffer, len);
+  struct sv rest;
+  u64 volume = sv_parse_unsigned_number(string_view, &rest);
+  int i = sv_length(string_view) - sv_length(rest);
   if (0 == i) {
     return 0;
   }
+
+  if (volume > 100) {
+    volume = 0;
+  }
+
   ac97_set_volume(volume);
   return i;
 }
@@ -52,18 +47,8 @@ int volume_read(u8 *buffer, u64 offset, u64 len, vfs_fd_t *fd) {
     return 0;
   }
   (void)fd;
-  if (len < 3) {
-    return 0;
-  }
   int volume = ac97_get_volume();
-  assert(volume <= 100);
-  if (volume == 100) {
-    memcpy(buffer, "100", 3);
-    return 3;
-  }
-  buffer[1] = (volume % 10) + '0';
-  buffer[0] = ((volume - (volume % 10)) / 10) + '0';
-  return 2;
+  return min(len, (u64)kbnprintf(buffer, len, "%d", volume));
 }
 
 static int add_files(void) {
diff --git a/kernel/includes/ctype.h b/kernel/includes/ctype.h
index f6be4f2..e8e0bae 100644
--- a/kernel/includes/ctype.h
+++ b/kernel/includes/ctype.h
@@ -1,2 +1,3 @@
 int isdigit(int c);
 int tolower(int c);
+int isspace(int c);
diff --git a/kernel/lib/sb.c b/kernel/lib/sb.c
new file mode 100644
index 0000000..3f99d95
--- /dev/null
+++ b/kernel/lib/sb.c
@@ -0,0 +1,124 @@
+#include <kmalloc.h>
+#include <math.h>
+#include <stdlib.h>
+#include <string.h>
+#include <tb/sb.h>
+
+#define DEFAULT_CAPACITY 256
+
+void sb_init(struct sb *ctx) {
+  (void)sb_init_capacity(ctx, DEFAULT_CAPACITY);
+}
+
+int sb_init_capacity(struct sb *ctx, size_t starting_capacity) {
+  ctx->length = 0;
+  ctx->prebuffer = 0;
+  ctx->string = kmalloc(starting_capacity);
+  if (NULL == ctx->string) {
+    ctx->capacity = 0;
+    return 0;
+  }
+  ctx->capacity = starting_capacity;
+  return 1;
+}
+
+void sb_init_buffer(struct sb *ctx, char *buffer, size_t size) {
+  ctx->string = buffer;
+  ctx->capacity = size;
+  ctx->length = 0;
+  ctx->prebuffer = 1;
+}
+
+void sb_free(struct sb *ctx) {
+  if (ctx->prebuffer) {
+    ctx->length = 0;
+    return;
+  }
+  ctx->length = 0;
+  ctx->capacity = 0;
+  kfree(ctx->string);
+  ctx->string = NULL;
+}
+
+void sb_reset(struct sb *ctx) {
+  ctx->length = 0;
+}
+
+int sb_isempty(const struct sb *ctx) {
+  return (0 == ctx->length);
+}
+
+int sb_increase_buffer(struct sb *ctx, size_t min) {
+  if (ctx->prebuffer) {
+    return 0;
+  }
+  size_t new_capacity = ctx->capacity + max(32, min);
+  char *new_allocation = krealloc(ctx->string, new_capacity);
+  if (!new_allocation) {
+    return 0;
+  }
+
+  ctx->string = new_allocation;
+  ctx->capacity = new_capacity;
+  return 1;
+}
+
+int sb_append_char(struct sb *ctx, char c) {
+  if (1 > ctx->capacity - ctx->length) {
+    if (!sb_increase_buffer(ctx, 1)) {
+      return 0;
+    }
+  }
+  memcpy(ctx->string + ctx->length, &c, 1);
+  ctx->length++;
+  return 1;
+}
+
+int sb_delete_right(struct sb *ctx, size_t n) {
+  n = min(n, ctx->length);
+  ctx->length -= n;
+  return n;
+}
+
+int sb_append(struct sb *ctx, const char *s) {
+  size_t l = strlen(s);
+  if (l > ctx->capacity - ctx->length) {
+    if (!sb_increase_buffer(ctx, l)) {
+      return 0;
+    }
+  }
+  memcpy(ctx->string + ctx->length, s, l);
+  ctx->length += l;
+  return 1;
+}
+
+int sb_prepend_sv(struct sb *ctx, struct sv sv) {
+  return sb_prepend_buffer(ctx, sv.s, sv.length);
+}
+
+int sb_prepend_buffer(struct sb *ctx, const char *buffer, size_t length) {
+  if (length > ctx->capacity - ctx->length) {
+    if (!sb_increase_buffer(ctx, length)) {
+      return 0;
+    }
+  }
+  memmove(ctx->string + length, ctx->string, ctx->length);
+  memcpy(ctx->string, buffer, length);
+  ctx->length += length;
+  return 1;
+}
+
+int sb_append_buffer(struct sb *ctx, const char *buffer, size_t length) {
+  if (length > ctx->capacity - ctx->length) {
+    if (!sb_increase_buffer(ctx, length)) {
+      return 0;
+    }
+  }
+  memcpy(ctx->string + ctx->length, buffer, length);
+  ctx->length += length;
+  return 1;
+}
+
+int sb_append_sv(struct sb *ctx, struct sv sv) {
+  return sb_append_buffer(ctx, sv.s, sv.length);
+}
diff --git a/kernel/lib/sb.h b/kernel/lib/sb.h
new file mode 100644
index 0000000..8493997
--- /dev/null
+++ b/kernel/lib/sb.h
@@ -0,0 +1,30 @@
+#ifndef SB_H
+#define SB_H
+#include "sv.h"
+#include <stddef.h>
+#include <stdint.h>
+
+struct sb {
+  char *string;
+  size_t length;
+  size_t capacity;
+  uint8_t prebuffer;
+};
+
+struct sv;
+
+void sb_init(struct sb *ctx);
+int sb_init_capacity(struct sb *ctx, size_t starting_capacity);
+void sb_init_buffer(struct sb *ctx, char *buffer, size_t size);
+void sb_free(struct sb *ctx);
+void sb_reset(struct sb *ctx);
+int sb_isempty(const struct sb *ctx);
+int sb_append_char(struct sb *ctx, char c);
+int sb_delete_right(struct sb *ctx, size_t n);
+int sb_append(struct sb *ctx, const char *s);
+int sb_append_buffer(struct sb *ctx, const char *buffer,
+                      size_t length);
+int sb_append_sv(struct sb *ctx, struct sv sv);
+int sb_prepend_sv(struct sb *ctx, struct sv sv);
+int sb_prepend_buffer(struct sb *ctx, const char *buffer, size_t length);
+#endif
diff --git a/kernel/lib/sv.c b/kernel/lib/sv.c
new file mode 100644
index 0000000..580e63a
--- /dev/null
+++ b/kernel/lib/sv.c
@@ -0,0 +1,247 @@
+#include <ctype.h>
+#include <kmalloc.h>
+#include <lib/sv.h>
+#include <math.h>
+#include <stdlib.h>
+#include <string.h>
+
+struct sv sv_init(const char *s, size_t length) {
+  return (struct sv){.s = s, .length = length};
+}
+
+char *SV_TO_C(struct sv s) {
+  char *c_string = kmalloc(s.length + 1);
+  memcpy(c_string, s.s, s.length);
+  c_string[s.length] = '\0';
+  return c_string;
+}
+
+size_t sv_to_cstring_buffer(struct sv s, char *buffer, size_t length) {
+  if (0 == length || length - 1 < s.length) {
+    return s.length;
+  }
+  memcpy(buffer, s.s, s.length);
+  buffer[s.length] = '\0';
+  return s.length;
+}
+
+struct sv sv_next(struct sv s) {
+  if (0 == s.length) {
+    return s;
+  }
+  s.length--;
+  s.s++;
+  return s;
+}
+
+struct sv sv_skip_chars(const struct sv input, const char *chars) {
+  struct sv r = input;
+  for (; r.length > 0;) {
+    int found = 0;
+    const char *p = chars;
+    for (; *p; p++) {
+      if (*p == r.s[0]) {
+        r.s++;
+        r.length--;
+        found = 1;
+        break;
+      }
+    }
+    if (!found) {
+      break;
+    }
+  }
+  return r;
+}
+
+uint64_t sv_parse_unsigned_number(struct sv input, struct sv *rest) {
+  uint64_t r = 0;
+  size_t i = 0;
+  for (; i < input.length; i++) {
+    if (!isdigit(input.s[i])) {
+      break;
+    }
+    r *= 10;
+    r += input.s[i] - '0';
+  }
+  input.length -= i;
+  input.s += i;
+  if (rest) {
+    *rest = input;
+  }
+  return r;
+}
+
+struct sv sv_split_function(const struct sv input, struct sv *rest,
+                            int (*function)(int)) {
+  struct sv r = {
+      .s = input.s,
+  };
+  for (size_t i = 0; i < input.length; i++) {
+    if (function(input.s[i])) {
+      r.length = i;
+      if (rest) {
+        rest->s += i;
+        rest->length -= i;
+      }
+      return r;
+    }
+  }
+
+  if (rest) {
+    rest->s = NULL;
+    rest->length = 0;
+  }
+  return input;
+}
+
+struct sv sv_split_space(const struct sv input, struct sv *rest) {
+  return sv_split_function(input, rest, isspace);
+}
+
+struct sv sv_end_split_delim(const struct sv input, struct sv *rest,
+                             char delim) {
+  for (size_t i = input.length - 1; i > 0; i--) {
+    if (delim == input.s[i]) {
+      struct sv r = {
+          .s = (input.s + i),
+          .length = input.length - i,
+      };
+      if (rest) {
+        rest->s = input.s;
+        rest->length = i;
+      }
+      return r;
+    }
+  }
+
+  if (rest) {
+    rest->s = NULL;
+    rest->length = 0;
+  }
+  return input;
+}
+
+struct sv sv_split_delim(const struct sv input, struct sv *rest, char delim) {
+  struct sv r = {
+      .s = input.s,
+  };
+  for (size_t i = 0; i < input.length; i++) {
+    if (delim == input.s[i]) {
+      r.length = i;
+      if (rest) {
+        rest->s += i + 1;
+        rest->length -= (i + 1);
+      }
+      return r;
+    }
+  }
+
+  if (rest) {
+    rest->s = NULL;
+    rest->length = 0;
+  }
+  return input;
+}
+
+int sv_isempty(struct sv s) {
+  return (0 == s.length);
+}
+
+char sv_peek(struct sv s) {
+  if (0 == s.length) {
+    return '\0';
+  }
+  return s.s[0];
+}
+
+int sv_partial_eq(struct sv a, struct sv b) {
+  if (a.length < b.length) {
+    return 0;
+  }
+  for (size_t i = 0; i < b.length; i++) {
+    if (a.s[i] != b.s[i]) {
+      return 0;
+    }
+  }
+  return 1;
+}
+
+int sv_eq(struct sv a, struct sv b) {
+  if (a.length != b.length) {
+    return 0;
+  }
+  for (size_t i = 0; i < a.length; i++) {
+    if (a.s[i] != b.s[i]) {
+      return 0;
+    }
+  }
+  return 1;
+}
+
+struct sv sv_take(struct sv s, struct sv *rest, size_t n) {
+  if (s.length < n) {
+    if (rest) {
+      rest->length = 0;
+    }
+    return s;
+  }
+  s.length = n;
+  if (rest) {
+    rest->length -= n;
+    rest->s += n;
+  }
+  return s;
+}
+
+struct sv sv_take_end(struct sv s, struct sv *rest, size_t n) {
+  if (s.length < n) {
+    if (rest) {
+      rest->length = 0;
+    }
+    return s;
+  }
+  if (rest) {
+    rest->length = s.length - n;
+  }
+  s.s += (s.length - n);
+  s.length = n;
+  return s;
+}
+
+struct sv sv_trim_left(struct sv s, size_t n) {
+  if (s.length < n) {
+    s.s += s.length;
+    s.length = 0;
+    return s;
+  }
+  s.s += n;
+  s.length -= n;
+  return s;
+}
+
+struct sv sv_clone(struct sv s) {
+  struct sv new_sv;
+  new_sv.length = s.length;
+  char *new_string = kmalloc(s.length);
+  memcpy(new_string, s.s, s.length);
+  new_sv.s = new_string;
+  return new_sv;
+}
+
+char *sv_copy_to_c(struct sv s, char *out, size_t buffer_length) {
+  int copy_len = min(s.length + 1, buffer_length);
+  if (0 == copy_len) {
+    return NULL;
+  }
+  if (!out) {
+    out = kmalloc(copy_len);
+  }
+  memcpy(out, s.s, copy_len - 1);
+  out[copy_len - 1] = '\0';
+  return out;
+}
+
+struct sv sv_clone_from_c(const char *s) {
+  return sv_clone(C_TO_SV(s));
+}
diff --git a/kernel/lib/sv.h b/kernel/lib/sv.h
new file mode 100644
index 0000000..b5cb0c9
--- /dev/null
+++ b/kernel/lib/sv.h
@@ -0,0 +1,43 @@
+#ifndef SV_H
+#define SV_H
+#include "sb.h"
+#include <stddef.h>
+
+#define SB_TO_SV(_sb)                                                          \
+  (struct sv) {                                                                \
+    .s = (_sb).string, .length = (_sb).length                                  \
+  }
+
+#define C_TO_SV(_c_string)                                                     \
+  ((struct sv){.length = strlen(_c_string), .s = (_c_string)})
+
+#define sv_length(a) ((a).length)
+
+struct sv {
+  const char *s;
+  size_t length;
+};
+
+struct sv sv_init(const char *s, size_t length);
+char *SV_TO_C(struct sv s);
+size_t sv_to_cstring_buffer(struct sv s, char *buffer, size_t length);
+struct sv sv_split_delim(const struct sv input, struct sv *rest, char delim);
+struct sv sv_end_split_delim(const struct sv input, struct sv *rest,
+                             char delim);
+struct sv sv_split_space(const struct sv input, struct sv *rest);
+struct sv sv_skip_chars(const struct sv input, const char *chars);
+struct sv sv_split_function(const struct sv input, struct sv *rest,
+                            int (*function)(int));
+struct sv sv_take(struct sv s, struct sv *rest, size_t n);
+struct sv sv_take_end(struct sv s, struct sv *rest, size_t n);
+struct sv sv_next(struct sv s);
+int sv_isempty(struct sv s);
+char sv_peek(struct sv s);
+int sv_eq(struct sv a, struct sv b);
+int sv_partial_eq(struct sv a, struct sv b);
+struct sv sv_trim_left(struct sv s, size_t n);
+struct sv sv_clone(struct sv s);
+struct sv sv_clone_from_c(const char *s);
+char *sv_copy_to_c(struct sv s, char *out, size_t buffer_length);
+uint64_t sv_parse_unsigned_number(struct sv input, struct sv *rest);
+#endif
diff --git a/kernel/libc/ctype/isspace.c b/kernel/libc/ctype/isspace.c
new file mode 100644
index 0000000..a922d91
--- /dev/null
+++ b/kernel/libc/ctype/isspace.c
@@ -0,0 +1,5 @@
+#include <ctype.h>
+
+int isspace(int c) {
+  return c == ' ' || (unsigned)c - '\t' < 5;
+}
diff --git a/kernel/libc/include/string.h b/kernel/libc/include/string.h
index 89ea5ab..4425085 100644
--- a/kernel/libc/include/string.h
+++ b/kernel/libc/include/string.h
@@ -15,4 +15,5 @@ char *copy_and_allocate_string(const char *s);
 char *copy_and_allocate_user_string(const char *s);
 size_t strlcpy(char *dst, const char *src, size_t dsize);
 char *strcat(char *s1, const char *s2);
+void *memmove(void *s1, const void *s2, size_t n);
 #endif
diff --git a/kernel/libc/string/memmove.c b/kernel/libc/string/memmove.c
new file mode 100644
index 0000000..5fc49f7
--- /dev/null
+++ b/kernel/libc/string/memmove.c
@@ -0,0 +1,14 @@
+#include <string.h>
+
+// copy bytes in memory with overlapping areas
+// https://pubs.opengroup.org/onlinepubs/9699919799/functions/memmove.html
+void *memmove(void *s1, const void *s2, size_t n) {
+  // Copying takes place as if the n bytes from the object pointed to by s2 are
+  // first copied into a temporary array of n bytes that does not overlap the
+  // objects pointed to by s1 and s2, and then the n bytes from the temporary
+  // array are copied into the object pointed to by s1.
+  unsigned char tmp[n];
+  memcpy(tmp, s2, n);
+  memcpy(s1, tmp, n);
+  return s1;
+}
diff --git a/kernel/timer.c b/kernel/timer.c
index dd92d3d..b84c16a 100644
--- a/kernel/timer.c
+++ b/kernel/timer.c
@@ -2,6 +2,7 @@
 #include <drivers/cmos.h>
 #include <fs/devfs.h>
 #include <interrupts.h>
+#include <lib/sv.h>
 #include <math.h>
 #include <random.h>
 #include <time.h>
@@ -49,31 +50,18 @@ int clock_read(u8 *buffer, u64 offset, u64 len, vfs_fd_t *fd) {
   if (0 != offset) {
     return 0;
   }
-  // Currently don't have snprintf without NULL termination...
-  char tmp[4096];
   u64 r = timer_get_ms();
-
-  u64 l = ksnprintf(tmp, 4096, "%llu", r);
-  l = min(len, l);
-  memcpy(buffer, tmp, l);
-  return l;
+  return min(len, (u64)kbnprintf(buffer, len, "%llu", r));
 }
 
 int clock_write(u8 *buffer, u64 offset, u64 len, vfs_fd_t *fd) {
   if (0 != offset) {
     return 0;
   }
-  // TODO: Move sv to kernel or something to just avoid this. I hate null
-  // termination
-  char tmp[4096];
-  memcpy(tmp, buffer, len);
-  tmp[len] = '\0';
 
-  int err;
-  u64 new_value_ms = parse_u64(tmp, NULL, 10, &err);
-  if (err) {
-    return 0;
-  }
+  struct sv string_view = sv_init(buffer, len);
+
+  u64 new_value_ms = sv_parse_unsigned_number(string_view, &string_view);
 
   i64 new_value_seconds = new_value_ms / 1000;