Навеянно постом Стань Ъ Линуксоидом....
Давно слышал интересные вещи про комментарии в коде ядра, решил поковырять, и не был разочарован. Итак, начнем.

./net/ipv4/tcp_input.c:701: * all the algo is pure shit and should be replaced

1 2 3 4 5 6 7 8 9

/* 2. Fixups made earlier cannot be right.          *    If we do not estimate RTO correctly without them,          *    all the algo is pure shit and should be replaced          *    with correct one. It is exactly, which we pretend to do.          */

./drivers/net/sunhme.c:930: /* Remember: "Different name, same old buggy as shit hardware." */

1 2 3

/* Remember: "Different name, same old buggy as shit hardware." */         HMD(("done\n"));

./drivers/net/wireless/iwlwifi/iwl3945-base.c:7594: /* all this shit doesn't belong into sysfs anyway */


./drivers/scsi/qlogicpti.c:782: if (qpti->clock == 0) /* bullshit */

1 2 3

if (qpti->clock == 0) /* bullshit */                 qpti->clock = 40;

./drivers/ata/sata_via.c:316: * SCR registers on vt6420 are pieces of shit and may hang the

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39

/**  *      vt6420_prereset - prereset for vt6420  *      @link: target ATA link  *      @deadline: deadline jiffies for the operation  *  *      SCR registers on vt6420 are pieces of shit and may hang the  *      whole machine completely if accessed with the wrong timing.  *      To avoid such catastrophe, vt6420 doesn't provide generic SCR  *      access operations, but uses SStatus and SControl only during  *      boot probing in controlled way.  *  *      As the old (pre EH update) probing code is proven to work, we  *      strictly follow the access pattern.  *  *      LOCKING:  *      Kernel thread context (may sleep)  *  *      RETURNS:  *      0 on success, -errno otherwise.  */

./drivers/mmc/host/sdhci.c:1063: * Some (ENE) controllers go apeshit on some ios operation,

1 2 3 4 5 6 7 8 9

/*          * Some (ENE) controllers go apeshit on some ios operation,          * signalling timeout and CRC errors even on CMD0. Resetting          * it on each ios seems to solve the problem.          */

./arch/sparc/kernel/pcic.c:497: * to shit into regions like that.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

/*                                  * A device responds to I/O cycles on PCI.                                  * We generate these cycles with memory                                  * access into the fixed map (phys 0x30000000).                                  *                                  * Since a device driver does not want to                                  * do ioremap() before accessing PC-style I/O,                                  * we supply virtual, ready to access address.                                  *                                  * Note that request_region()                                  * works for these devices.                                  *                                  * XXX Neat trick, but it's a *bad* idea                                  * to shit into regions like that.                                  * What if we want to allocate one more                                  * PCI base address...                                  */

./arch/sparc/mm/srmmu.c:1607: * this shit off... nice job Fujitsu.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

/*          * The Swift branch folding logic is completely broken.  At          * trap time, if things are just right, if can mistakenly          * think that a trap is coming from kernel mode when in fact          * it is coming from user mode (it mis-executes the branch in          * the trap code).  So you see things like crashme completely          * hosing your machine which is completely unacceptable.  Turn          * this shit off... nice job Fujitsu.          */

./sound/pci/ac97/ac97_patch.c:3612: * shit may happen.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

/*  * ...  * I have no idea what the hell Reserved does, but on an MSI  * CN700T, i have to set it to get 5.1 output - YMMV, bad  * shit may happen.  *  * If other chips use Universal Audio Jack, then this code might be applicable  * to them.  * ... */

./fs/xfs/xfs_btree.h:137: case XFS_BTNUM_MAX: ASSERT(0); /* fucking gcc */ ; break; \

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

#define XFS_BTREE_STATS_INC(cur, stat)  \ do {    \         switch (cur->bc_btnum) {  \         case XFS_BTNUM_BNO: __XFS_BTREE_STATS_INC(abtb, stat); break;   \         case XFS_BTNUM_CNT: __XFS_BTREE_STATS_INC(abtc, stat); break;   \         case XFS_BTNUM_BMAP: __XFS_BTREE_STATS_INC(bmbt, stat); break;  \         case XFS_BTNUM_INO: __XFS_BTREE_STATS_INC(ibt, stat); break;    \         case XFS_BTNUM_MAX: ASSERT(0); /* fucking gcc */ ; break;       \         }       \ } while (0)

./linux-2.6.29-gentoo-r5/net/ipv4/netfilter/nf_nat_snmp_basic.c::1030 * (And this is the fucking 'basic' method).

1 2 3 4 5 6 7

/*  * Parse and mangle SNMP message according to mapping.  * (And this is the fucking 'basic' method).  */

./linux-2.6.29-gentoo-r5/lib/vsprintf.c:9: * Wirzenius wrote this portably, Torvalds fucked it up :-)

1 2 3 4 5 6 7 8 9

/* vsprintf.c -- Lars Wirzenius & Linus Torvalds. */ /*  * Wirzenius wrote this portably, Torvalds fucked it up :-) ... */

./linux-2.6.29-gentoo-r5/drivers/net/sunhme.c:986:/* Only Sun can take such nice parts and fuck up the programming interface

1 2 3 4 5 6 7

/* Only Sun can take such nice parts and fuck up the programming interface  * like this.  Good job guys...  */

./linux-2.6.29-gentoo-r5/drivers/net/sunhme.c:2067: /* This card is _fucking_ hot... */

./linux-2.6.29-gentoo-r5/drivers/scsi/qlogicpti.h:64:/* Am I fucking pedantic or what? */

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

/* Am I fucking pedantic or what? */ struct Entry_header { #ifdef __BIG_ENDIAN         u8      entry_cnt;         u8      entry_type;         u8      flags;         u8      sys_def_1; #else /* __LITTLE_ENDIAN */         u8      entry_type;         u8      entry_cnt;         u8      sys_def_1;         u8      flags; #endif };

./linux-2.6.29-gentoo-r5/arch/x86/kernel/cpu/cpufreq/powernow-k7.c:585: * Some Athlon laptops have really fucked PST tables.

1 2 3 4 5 6 7 8 9

/*  * Some Athlon laptops have really fucked PST tables.  * A BIOS update is all that can save them.  * Mention this, and disable cpufreq.  */

./linux-2.6.29-gentoo-r5/arch/m68k/include/asm/sun3ints.h:30:/* master list of VME vectors -- don't fuck with this */

./linux-2.6.29-gentoo-r5/arch/sparc/include/asm/system_64.h:192: /* If you fuck with this, update ret_from_syscall code too. */

./linux-2.6.29-gentoo-r5/arch/mips/pci/ops-bridge.c:59: * IOC3 is fucked fucked beyond believe ... Don't even give the

1 2 3 4 5 6 7

/*          * IOC3 is fucked fucked beyond believe ...  Don't even give the          * generic PCI code a chance to look at the wrong register.          */

./linux-2.6.29-gentoo-r5/sound/oss/opl3.c:833: * What the fuck is going on here? We leave junk in the beginning

1 2 3 4 5 6 7

/*          * What the fuck is going on here?  We leave junk in the beginning          * of ins and then check the field pretty close to that beginning?          */

./linux-2.6.29-gentoo-r5/net/irda/irnet/irnet_irda.c:1221: /* Give a kick in the ass of ppp_generic so that he sends us some data */

1 2 3 4 5

/* Give a kick in the ass of ppp_generic so that he sends us some data */   ppp_output_wakeup(&self->chan);

./linux-2.6.29-gentoo-r5/arch/arm/mach-lh7a40x/include/mach/hardware.h:49:/* Let's kick gcc's ass again... */

1 2 3 4 5 6 7

/* Let's kick gcc's ass again... */ # define __REG2(x,y)    \         ( __builtin_constant_p(y) ? (__REG((x) + (y))) \                                   : (*(volatile u32 *)((u32)&__REG(x) + (y))) )

./linux-2.6.29-gentoo-r5/arch/mips/kernel/process.c:433: * get_wchan - a maintenance nightmare^W^Wpain in the ass ...

1 2 3 4 5

/*  * get_wchan - a maintenance nightmare^W^Wpain in the ass ...  */

PS: тут не все, но и этого думаю достаточно)