Bugs in the multiboot example kernel implementation
PATCH 1/1 Bugs in the multiboot example kernel implementation
I think there are two bugs in the example implementation of the multiboot standard version 0.6.96. Both concern the display of the memory map.
First, the `multiboot_uint64_t' type does not seem to be working as expected. I always get 32-bit there. I personally wouldn't use that type anyway, so I used two 32-bit variables in place.
Second, the memory addresses *must* be zero-padded when chaining the output of two `itoa' calls this way, so I introduced the (non-standard) '%X' specifier.
As this is the first time I tinker with GRUB code and also almost the first time I edit kernel code, it probably breaks lots of conventions... But I hope it can be turned into something useful.
For the legal stuff:
I hereby put this patch into the public domain; as I'm not sure if this is possible in Germany, I explicitly grant anyone the right to use this patch in any thinkable way, including removing any reference to me and relicensing, as a whole or in parts, and claiming copyright for it.
docs/kernel.c +27 -7
@@ -141,13 +141,13 @@ (unsigned long ) mmap < mbi->mmap_addr + mbi->mmap_length; mmap = (multiboot_memory_map_t *) ((unsigned long) mmap + mmap->size + sizeof (mmap->size))) -printf (" size = 0x%x, base_addr = 0x%x%x," -" length = 0x%x%x, type = 0x%x\n" , +printf (" size = 0x%x, base_addr = 0x%X%X," +" length = 0x%X%X, type = 0x%x\n" , (unsigned ) mmap->size, - mmap->addr >>32 , - mmap->addr &0xffffffff , - mmap->len >>32 , - mmap->len &0xffffffff , + mmap->addr1, + mmap->addr0, + mmap->len1, + mmap->len0, (unsigned ) mmap->type); } }
@@ -185,7 +185,7 @@ buf++; ud = -d; } -else if (base =='x' ) +else if (base =='x' || base =='X' ) divisor =16 ;/* Divide UD by DIVISOR until UD == 0. */
@@ -213,6 +213,23 @@ } } +static void +pad0 (char *numstr,int n) +{ +int i, len; + +/* strlen() */ +for (len =0 ; numstr[len]; len++); + +/* shift right */ +for (i = len -1 ; i >=0 ; i--) + numstr[i + n - len] = numstr[i]; + +/* pad */ +for (i =0 ; i < n - len; i++) + numstr[i] ='0' ; +} +/* Put the character C on the screen. */ static void putchar (int c)
@@ -260,8 +277,11 @@case 'd' :case 'u' :case 'x' : +case 'X' :itoa (buf, c, *((int *) arg++)); p = buf; +if (c =='X' ) +pad0 (p,8 ); goto string;break ;
diff -aur multiboot-0.6.96/docs/kernel.c.texi multiboot-0.6.96.new/docs/kernel.c.texi
--- multiboot-0.6.96/docs/kernel.c.texi 2009-12-24 15:23:11.000000000 +0100
+++ multiboot-0.6.96.new/docs/kernel.c.texi 2010-02-27
02:35:32.000000000 +0100
@@ -141,13 +141,13 @@
(unsigned long) mmap < mbi->mmap_addr + mbi->mmap_length; mmap = (multiboot_memory_map_t *) ((unsigned long) mmap
+ mmap->size + sizeof (mmap->size)))
- printf (" size = 0x%x, base_addr = 0x%x%x,"
- " length = 0x%x%x, type = 0x%x\n",
+ printf (" size = 0x%x, base_addr = 0x%X%X,"
+ " length = 0x%X%X, type = 0x%x\n",
(unsigned) mmap->size,
- mmap->addr >> 32,
- mmap->addr & 0xffffffff,
- mmap->len >> 32,
- mmap->len & 0xffffffff,
+ mmap->addr1,
+ mmap->addr0,
+ mmap->len1,
+ mmap->len0,
(unsigned) mmap->type);
@}
@}
@@ -185,7 +185,7 @@
buf++;
ud = -d;
@}
- else if (base == 'x')
+ else if (base == 'x' || base == 'X')
divisor = 16; /* @r{Divide UD by DIVISOR until UD == 0.} */
@@ -213,6 +213,23 @@
@}
@}
+static void
+pad0 (char *numstr, int n)
+@{
+ int i, len;
+
+ /* @r{strlen()} */
+ for (len = 0; numstr[len]; len++);
+
+ /* @r{shift right} */
+ for (i = len - 1; i >= 0; i--)
+ numstr[i + n - len] = numstr[i];
+
+ /* @r{pad} */
+ for (i = 0; i < n - len; i++)
+ numstr[i] = '0';
+@}
+
/* @r{Put the character C on the screen.} */
static void
putchar (int c)
@@ -260,8 +277,11 @@
case 'd':
case 'u':
case 'x':
+ case 'X':
itoa (buf, c, *((int *) arg++));
p = buf;
+ if (c == 'X')
+ pad0 (p, 8);
goto string;
break;diff -aur multiboot-0.6.96/docs/multiboot.hmultiboot-0.6.96.new/docs/multiboot.h
--- multiboot-0.6.96/docs/multiboot.h 2009-12-24 15:19:40.000000000 +0100
+++ multiboot-0.6.96.new/docs/multiboot.h 2010-02-27 01:50:47.000000000
+0100
@@ -196,12 +196,14 @@
struct multiboot_mmap_entry
{
multiboot_uint32_t size;
- multiboot_uint64_t addr; - multiboot_uint64_t len; + multiboot_uint32_t addr0; + multiboot_uint32_t addr1; + multiboot_uint32_t len0; + multiboot_uint32_t len1;
#define MULTIBOOT_MEMORY_AVAILABLE 1
#define MULTIBOOT_MEMORY_RESERVED 2
multiboot_uint32_t type;
-} __attribute__((packed));
+};
typedef struct multiboot_mmap_entry multiboot_memory_map_t;
struct multiboot_mod_list
diff -aur multiboot-0.6.96/docs/multiboot.h.texi
multiboot-0.6.96.new/docs/multiboot.h.texi
--- multiboot-0.6.96/docs/multiboot.h.texi 2009-12-24 15:19:41.000000000
+0100
+++ multiboot-0.6.96.new/docs/multiboot.h.texi 2010-02-27
01:51:10.000000000 +0100
@@ -196,12 +196,14 @@
struct multiboot_mmap_entry
@{
multiboot_uint32_t size;
- multiboot_uint64_t addr; - multiboot_uint64_t len; + multiboot_uint32_t addr0; + multiboot_uint32_t addr1; + multiboot_uint32_t len0; + multiboot_uint32_t len1;
#define MULTIBOOT_MEMORY_AVAILABLE 1
#define MULTIBOOT_MEMORY_RESERVED 2
multiboot_uint32_t type;
-@} __attribute__((packed));
+@};
typedef struct multiboot_mmap_entry multiboot_memory_map_t;
struct multiboot_mod_list
diff -aur multiboot-0.6.96/docs/multiboot.info multiboot-0.6.96.new/docs/multiboot.info
--- multiboot-0.6.96/docs/multiboot.info 2009-12-24 16:38:18.000000000 +0100
+++ multiboot-0.6.96.new/docs/multiboot.info 2010-02-27
02:35:32.000000000 +0100
@@ -1,4 +1,4 @@
-This is multiboot.info, produced by makeinfo version 4.11 from +This is multiboot.info, produced by makeinfo version 4.13 from
multiboot.texi.
Copyright (C) 1995,96 Bryan Ford <[email protected]>
@@ -1269,12 +1269,14 @@
struct multiboot_mmap_entry
{
multiboot_uint32_t size;
- multiboot_uint64_t addr;
- multiboot_uint64_t len;
+ multiboot_uint32_t addr0;
+ multiboot_uint32_t addr1;
+ multiboot_uint32_t len0;
+ multiboot_uint32_t len1;
#define MULTIBOOT_MEMORY_AVAILABLE 1
#define MULTIBOOT_MEMORY_RESERVED 2
multiboot_uint32_t type;
- } __attribute__((packed));
+ };
typedef struct multiboot_mmap_entry multiboot_memory_map_t;
struct multiboot_mod_list@@ -1551,13 +1553,13 @@
(unsigned long) mmap < mbi->mmap_addr + mbi->mmap_length; mmap = (multiboot_memory_map_t *) ((unsigned long) mmap
+ mmap->size + sizeof
(mmap->size)))
- printf (" size = 0x%x, base_addr = 0x%x%x,"
- " length = 0x%x%x, type = 0x%x\n",
+ printf (" size = 0x%x, base_addr = 0x%X%X,"
+ " length = 0x%X%X, type = 0x%x\n",
(unsigned) mmap->size,
- mmap->addr >> 32,
- mmap->addr & 0xffffffff,
- mmap->len >> 32,
- mmap->len & 0xffffffff,
+ mmap->addr1,
+ mmap->addr0,
+ mmap->len1,
+ mmap->len0,
(unsigned) mmap->type);
}
}
@@ -1595,7 +1597,7 @@
buf++;
ud = -d;
}
- else if (base == 'x')
+ else if (base == 'x' || base == 'X')
divisor = 16;
/* Divide UD by DIVISOR until UD == 0. */
@@ -1623,6 +1625,23 @@
}
}+ static void
+ pad0 (char *numstr, int n)
+ {+ int i, len;
+
+ /* strlen() */
+ for (len = 0; numstr[len]; len++);
+
+ /* shift right */
+ for (i = len - 1; i >= 0; i--)
+ numstr[i + n - len] = numstr[i];
+
+ /* pad */
+ for (i = 0; i < n - len; i++)
+ numstr[i] = '0';
+ }
+
/* Put the character C on the screen. */
static void
putchar (int c)
@@ -1670,8 +1689,11 @@
case 'd':
case 'u':
case 'x':
+ case 'X':
itoa (buf, c, *((int *) arg++));
p = buf;
+ if (c == 'X')
+ pad0 (p, 8);
goto string;
break;@@ -1800,11 +1822,11 @@
Node: I/O restriction technique42685
Node: Example OS code43902
Node: multiboot.h45444
-Node: boot.S53515
-Node: kernel.c56668
-Node: Other Multiboot kernels65537
-Node: Example boot loader code65968
-Node: History66494
-Node: Index67716
+Node: boot.S53558
+Node: kernel.c56711
+Node: Other Multiboot kernels65990
+Node: Example boot loader code66421
+Node: History66947
+Node: Index68169
End Tag Table
Quoted 9 lines · Soeren D. Schulze
I think there are two bugs in the example implementation of the multiboot standard version 0.6.96. Both concern the display of the memory map.
First, the `multiboot_uint64_t' type does not seem to be working as expected. I always get 32-bit there. I personally wouldn't use that type anyway, so I used two 32-bit variables in place.
This bug was fixed for some time in trunk
Quoted 4 lines
Second, the memory addresses *must* be zero-padded when chaining the output of two `itoa' calls this way, so I introduced the (non-standard) '%X' specifier.
Non-apropriate. I've just fixed it properly in bazaar. Thanks for reporting
Signature · 2 lines
Regards Vladimir 'φ-coder/phcoder' Serbinenko