linux/Documentation/printk-formats.txt
<<
>>
Prefs
   1If variable is of Type,         use printk format specifier:
   2---------------------------------------------------------
   3                int                     %d or %x
   4                unsigned int            %u or %x
   5                long                    %ld or %lx
   6                unsigned long           %lu or %lx
   7                long long               %lld or %llx
   8                unsigned long long      %llu or %llx
   9                size_t                  %zu or %zx
  10                ssize_t                 %zd or %zx
  11                s32                     %d or %x
  12                u32                     %u or %x
  13                s64                     %lld or %llx
  14                u64                     %llu or %llx
  15
  16If <type> is dependent on a config option for its size (e.g., sector_t,
  17blkcnt_t) or is architecture-dependent for its size (e.g., tcflag_t), use a
  18format specifier of its largest possible type and explicitly cast to it.
  19Example:
  20
  21        printk("test: sector number/total blocks: %llu/%llu\n",
  22                (unsigned long long)sector, (unsigned long long)blockcount);
  23
  24Reminder: sizeof() result is of type size_t.
  25
  26The kernel's printf does not support %n. For obvious reasons, floating
  27point formats (%e, %f, %g, %a) are also not recognized. Use of any
  28unsupported specifier or length qualifier results in a WARN and early
  29return from vsnprintf.
  30
  31Raw pointer value SHOULD be printed with %p. The kernel supports
  32the following extended format specifiers for pointer types:
  33
  34Symbols/Function Pointers:
  35
  36        %pF     versatile_init+0x0/0x110
  37        %pf     versatile_init
  38        %pS     versatile_init+0x0/0x110
  39        %pSR    versatile_init+0x9/0x110
  40                (with __builtin_extract_return_addr() translation)
  41        %ps     versatile_init
  42        %pB     prev_fn_of_versatile_init+0x88/0x88
  43
  44        For printing symbols and function pointers. The 'S' and 's' specifiers
  45        result in the symbol name with ('S') or without ('s') offsets. Where
  46        this is used on a kernel without KALLSYMS - the symbol address is
  47        printed instead.
  48
  49        The 'B' specifier results in the symbol name with offsets and should be
  50        used when printing stack backtraces. The specifier takes into
  51        consideration the effect of compiler optimisations which may occur
  52        when tail-call's are used and marked with the noreturn GCC attribute.
  53
  54        On ia64, ppc64 and parisc64 architectures function pointers are
  55        actually function descriptors which must first be resolved. The 'F' and
  56        'f' specifiers perform this resolution and then provide the same
  57        functionality as the 'S' and 's' specifiers.
  58
  59Kernel Pointers:
  60
  61        %pK     0x01234567 or 0x0123456789abcdef
  62
  63        For printing kernel pointers which should be hidden from unprivileged
  64        users. The behaviour of %pK depends on the kptr_restrict sysctl - see
  65        Documentation/sysctl/kernel.txt for more details.
  66
  67Struct Resources:
  68
  69        %pr     [mem 0x60000000-0x6fffffff flags 0x2200] or
  70                [mem 0x0000000060000000-0x000000006fffffff flags 0x2200]
  71        %pR     [mem 0x60000000-0x6fffffff pref] or
  72                [mem 0x0000000060000000-0x000000006fffffff pref]
  73
  74        For printing struct resources. The 'R' and 'r' specifiers result in a
  75        printed resource with ('R') or without ('r') a decoded flags member.
  76        Passed by reference.
  77
  78Physical addresses types phys_addr_t:
  79
  80        %pa[p]  0x01234567 or 0x0123456789abcdef
  81
  82        For printing a phys_addr_t type (and its derivatives, such as
  83        resource_size_t) which can vary based on build options, regardless of
  84        the width of the CPU data path. Passed by reference.
  85
  86DMA addresses types dma_addr_t:
  87
  88        %pad    0x01234567 or 0x0123456789abcdef
  89
  90        For printing a dma_addr_t type which can vary based on build options,
  91        regardless of the width of the CPU data path. Passed by reference.
  92
  93Raw buffer as an escaped string:
  94
  95        %*pE[achnops]
  96
  97        For printing raw buffer as an escaped string. For the following buffer
  98
  99                1b 62 20 5c 43 07 22 90 0d 5d
 100
 101        few examples show how the conversion would be done (the result string
 102        without surrounding quotes):
 103
 104                %*pE            "\eb \C\a"\220\r]"
 105                %*pEhp          "\x1bb \C\x07"\x90\x0d]"
 106                %*pEa           "\e\142\040\\\103\a\042\220\r\135"
 107
 108        The conversion rules are applied according to an optional combination
 109        of flags (see string_escape_mem() kernel documentation for the
 110        details):
 111                a - ESCAPE_ANY
 112                c - ESCAPE_SPECIAL
 113                h - ESCAPE_HEX
 114                n - ESCAPE_NULL
 115                o - ESCAPE_OCTAL
 116                p - ESCAPE_NP
 117                s - ESCAPE_SPACE
 118        By default ESCAPE_ANY_NP is used.
 119
 120        ESCAPE_ANY_NP is the sane choice for many cases, in particularly for
 121        printing SSIDs.
 122
 123        If field width is omitted the 1 byte only will be escaped.
 124
 125Raw buffer as a hex string:
 126
 127        %*ph    00 01 02  ...  3f
 128        %*phC   00:01:02: ... :3f
 129        %*phD   00-01-02- ... -3f
 130        %*phN   000102 ... 3f
 131
 132        For printing a small buffers (up to 64 bytes long) as a hex string with
 133        certain separator. For the larger buffers consider to use
 134        print_hex_dump().
 135
 136MAC/FDDI addresses:
 137
 138        %pM     00:01:02:03:04:05
 139        %pMR    05:04:03:02:01:00
 140        %pMF    00-01-02-03-04-05
 141        %pm     000102030405
 142        %pmR    050403020100
 143
 144        For printing 6-byte MAC/FDDI addresses in hex notation. The 'M' and 'm'
 145        specifiers result in a printed address with ('M') or without ('m') byte
 146        separators. The default byte separator is the colon (':').
 147
 148        Where FDDI addresses are concerned the 'F' specifier can be used after
 149        the 'M' specifier to use dash ('-') separators instead of the default
 150        separator.
 151
 152        For Bluetooth addresses the 'R' specifier shall be used after the 'M'
 153        specifier to use reversed byte order suitable for visual interpretation
 154        of Bluetooth addresses which are in the little endian order.
 155
 156        Passed by reference.
 157
 158IPv4 addresses:
 159
 160        %pI4    1.2.3.4
 161        %pi4    001.002.003.004
 162        %p[Ii]4[hnbl]
 163
 164        For printing IPv4 dot-separated decimal addresses. The 'I4' and 'i4'
 165        specifiers result in a printed address with ('i4') or without ('I4')
 166        leading zeros.
 167
 168        The additional 'h', 'n', 'b', and 'l' specifiers are used to specify
 169        host, network, big or little endian order addresses respectively. Where
 170        no specifier is provided the default network/big endian order is used.
 171
 172        Passed by reference.
 173
 174IPv6 addresses:
 175
 176        %pI6    0001:0002:0003:0004:0005:0006:0007:0008
 177        %pi6    00010002000300040005000600070008
 178        %pI6c   1:2:3:4:5:6:7:8
 179
 180        For printing IPv6 network-order 16-bit hex addresses. The 'I6' and 'i6'
 181        specifiers result in a printed address with ('I6') or without ('i6')
 182        colon-separators. Leading zeros are always used.
 183
 184        The additional 'c' specifier can be used with the 'I' specifier to
 185        print a compressed IPv6 address as described by
 186        http://tools.ietf.org/html/rfc5952
 187
 188        Passed by reference.
 189
 190IPv4/IPv6 addresses (generic, with port, flowinfo, scope):
 191
 192        %pIS    1.2.3.4         or 0001:0002:0003:0004:0005:0006:0007:0008
 193        %piS    001.002.003.004 or 00010002000300040005000600070008
 194        %pISc   1.2.3.4         or 1:2:3:4:5:6:7:8
 195        %pISpc  1.2.3.4:12345   or [1:2:3:4:5:6:7:8]:12345
 196        %p[Ii]S[pfschnbl]
 197
 198        For printing an IP address without the need to distinguish whether it's
 199        of type AF_INET or AF_INET6, a pointer to a valid 'struct sockaddr',
 200        specified through 'IS' or 'iS', can be passed to this format specifier.
 201
 202        The additional 'p', 'f', and 's' specifiers are used to specify port
 203        (IPv4, IPv6), flowinfo (IPv6) and scope (IPv6). Ports have a ':' prefix,
 204        flowinfo a '/' and scope a '%', each followed by the actual value.
 205
 206        In case of an IPv6 address the compressed IPv6 address as described by
 207        http://tools.ietf.org/html/rfc5952 is being used if the additional
 208        specifier 'c' is given. The IPv6 address is surrounded by '[', ']' in
 209        case of additional specifiers 'p', 'f' or 's' as suggested by
 210        https://tools.ietf.org/html/draft-ietf-6man-text-addr-representation-07
 211
 212        In case of IPv4 addresses, the additional 'h', 'n', 'b', and 'l'
 213        specifiers can be used as well and are ignored in case of an IPv6
 214        address.
 215
 216        Passed by reference.
 217
 218        Further examples:
 219
 220        %pISfc          1.2.3.4         or [1:2:3:4:5:6:7:8]/123456789
 221        %pISsc          1.2.3.4         or [1:2:3:4:5:6:7:8]%1234567890
 222        %pISpfc         1.2.3.4:12345   or [1:2:3:4:5:6:7:8]:12345/123456789
 223
 224UUID/GUID addresses:
 225
 226        %pUb    00010203-0405-0607-0809-0a0b0c0d0e0f
 227        %pUB    00010203-0405-0607-0809-0A0B0C0D0E0F
 228        %pUl    03020100-0504-0706-0809-0a0b0c0e0e0f
 229        %pUL    03020100-0504-0706-0809-0A0B0C0E0E0F
 230
 231        For printing 16-byte UUID/GUIDs addresses. The additional 'l', 'L',
 232        'b' and 'B' specifiers are used to specify a little endian order in
 233        lower ('l') or upper case ('L') hex characters - and big endian order
 234        in lower ('b') or upper case ('B') hex characters.
 235
 236        Where no additional specifiers are used the default big endian
 237        order with lower case hex characters will be printed.
 238
 239        Passed by reference.
 240
 241dentry names:
 242
 243        %pd{,2,3,4}
 244        %pD{,2,3,4}
 245
 246        For printing dentry name; if we race with d_move(), the name might be
 247        a mix of old and new ones, but it won't oops.  %pd dentry is a safer
 248        equivalent of %s dentry->d_name.name we used to use, %pd<n> prints
 249        n last components.  %pD does the same thing for struct file.
 250
 251        Passed by reference.
 252
 253block_device names:
 254
 255        %pg     sda, sda1 or loop0p1
 256
 257        For printing name of block_device pointers.
 258
 259struct va_format:
 260
 261        %pV
 262
 263        For printing struct va_format structures. These contain a format string
 264        and va_list as follows:
 265
 266        struct va_format {
 267                const char *fmt;
 268                va_list *va;
 269        };
 270
 271        Implements a "recursive vsnprintf".
 272
 273        Do not use this feature without some mechanism to verify the
 274        correctness of the format string and va_list arguments.
 275
 276        Passed by reference.
 277
 278struct clk:
 279
 280        %pC     pll1
 281        %pCn    pll1
 282        %pCr    1560000000
 283
 284        For printing struct clk structures. '%pC' and '%pCn' print the name
 285        (Common Clock Framework) or address (legacy clock framework) of the
 286        structure; '%pCr' prints the current clock rate.
 287
 288        Passed by reference.
 289
 290bitmap and its derivatives such as cpumask and nodemask:
 291
 292        %*pb    0779
 293        %*pbl   0,3-6,8-10
 294
 295        For printing bitmap and its derivatives such as cpumask and nodemask,
 296        %*pb output the bitmap with field width as the number of bits and %*pbl
 297        output the bitmap as range list with field width as the number of bits.
 298
 299        Passed by reference.
 300
 301Flags bitfields such as page flags, gfp_flags:
 302
 303        %pGp    referenced|uptodate|lru|active|private
 304        %pGg    GFP_USER|GFP_DMA32|GFP_NOWARN
 305        %pGv    read|exec|mayread|maywrite|mayexec|denywrite
 306
 307        For printing flags bitfields as a collection of symbolic constants that
 308        would construct the value. The type of flags is given by the third
 309        character. Currently supported are [p]age flags, [v]ma_flags (both
 310        expect unsigned long *) and [g]fp_flags (expects gfp_t *). The flag
 311        names and print order depends on the particular type.
 312
 313        Note that this format should not be used directly in TP_printk() part
 314        of a tracepoint. Instead, use the show_*_flags() functions from
 315        <trace/events/mmflags.h>.
 316
 317        Passed by reference.
 318
 319Network device features:
 320
 321        %pNF    0x000000000000c000
 322
 323        For printing netdev_features_t.
 324
 325        Passed by reference.
 326
 327If you add other %p extensions, please extend lib/test_printf.c with
 328one or more test cases, if at all feasible.
 329
 330
 331Thank you for your cooperation and attention.
 332
 333
 334By Randy Dunlap <rdunlap@infradead.org> and
 335Andrew Murray <amurray@mpc-data.co.uk>
 336