Skip to contentSkip to conversation

2.6.14 assorted warnings

DISCUSSION Keith Owens 9 participants 14 messages Oct 28, 2005 – Oct 31, 2005 VFS

lore
Share & tools
lore
#0 lore raw

gcc 4.0.1 on i386, 2.6.14 assorted warnings.

arch/i386/kernel/cpu/transmeta.c: In function 'init_transmeta':
arch/i386/kernel/cpu/transmeta.c:11: warning: 'cpu_freq' may be used uninitialized in this function

kernel/kmod.c: In function '____call_usermodehelper':
kernel/kmod.c:138: warning: statement with no effect

fs/bio.c: In function 'bio_alloc_bioset':
fs/bio.c:167: warning: 'idx' may be used uninitialized in this function

fs/nfsd/nfsctl.c: In function 'write_filehandle':
fs/nfsd/nfsctl.c:262: warning: 'maxsize' may be used uninitialized in this function

fs/udf/balloc.c: In function 'udf_table_new_block':
fs/udf/balloc.c:757: warning: 'goal_eloc.logicalBlockNum' may be used uninitialized in this function

fs/udf/super.c: In function 'udf_load_partition':
fs/udf/super.c:1351: warning: 'ino.partitionReferenceNum' may be used uninitialized in this function

fs/xfs/xfs_alloc_btree.c: In function 'xfs_alloc_insrec':
fs/xfs/xfs_alloc_btree.c:622: warning: 'nrec.ar_startblock' may be used uninitialized in this function fs/xfs/xfs_alloc_btree.c:622: warning: 'nrec.ar_blockcount' may be used uninitialized in this function

fs/xfs/xfs_bmap.c: In function 'xfs_bmap_alloc':
fs/xfs/xfs_bmap.c:2335: warning: 'rtx' is used uninitialized in this function

fs/xfs/xfs_dir2_sf.c: In function 'xfs_dir2_block_sfsize':
fs/xfs/xfs_dir2_sf.c:110: warning: 'parent' may be used uninitialized in this function

fs/xfs/xfs_dir_leaf.c: In function 'xfs_dir_leaf_to_shortform':
fs/xfs/xfs_dir_leaf.c:653: warning: 'parent' may be used uninitialized in this function

fs/xfs/xfs_ialloc_btree.c: In function 'xfs_inobt_insrec':
fs/xfs/xfs_ialloc_btree.c:750: warning: 'nrec.ir_free' is used uninitialized in this function fs/xfs/xfs_ialloc_btree.c:750: warning: 'nrec.ir_freecount' is used uninitialized in this function fs/xfs/xfs_ialloc_btree.c:567: warning: 'nrec.ir_startino' may be used uninitialized in this function

ipc/msg.c: In function 'sys_msgctl':
ipc/msg.c:333: warning: 'setbuf.qbytes' may be used uninitialized in this function ipc/msg.c:333: warning: 'setbuf.uid' may be used uninitialized in this function ipc/msg.c:333: warning: 'setbuf.gid' may be used uninitialized in this function ipc/msg.c:333: warning: 'setbuf.mode' may be used uninitialized in this function

ipc/sem.c: In function 'semctl_down':
ipc/sem.c:803: warning: 'setbuf.uid' may be used uninitialized in this function ipc/sem.c:803: warning: 'setbuf.gid' may be used uninitialized in this function ipc/sem.c:803: warning: 'setbuf.mode' may be used uninitialized in this function

drivers/net/wireless/ipw2100.c: In function 'store_memory':
drivers/net/wireless/ipw2100.c:3756: warning: unused variable 'dev' drivers/net/wireless/ipw2100.c: In function 'store_scan_age':
drivers/net/wireless/ipw2100.c:4065: warning: unused variable 'dev'

sound/core/oss/route.c: In function 'route_to_channel':
sound/core/oss/route.c:207: warning: 'src' may be used uninitialized in this function

lib/zlib_inflate/inftrees.c: In function 'huft_build':
lib/zlib_inflate/inftrees.c:121: warning: 'r.base' may be used uninitialized in this function

Stripped .config.

X86=y
SEMAPHORE_SLEEPERS=y
MMU=y
UID16=y
GENERIC_ISA_DMA=y
GENERIC_IOMAP=y
ARCH_MAY_HAVE_PC_FDC=y
EXPERIMENTAL=y
CLEAN_COMPILE=y
LOCK_KERNEL=y
INIT_ENV_ARG_LIMIT=32
LOCALVERSION="kaos"
SWAP=y
SYSVIPC=y
SYSCTL=y
HOTPLUG=y
KOBJECT_UEVENT=y
INITRAMFS_SOURCE=""
KALLSYMS=y
KALLSYMS_ALL=y
PRINTK=y
BUG=y
BASE_FULL=y
FUTEX=y
EPOLL=y
SHMEM=y
CC_ALIGN_FUNCTIONS=0
CC_ALIGN_LABELS=0
CC_ALIGN_LOOPS=0
CC_ALIGN_JUMPS=0
BASE_SMALL=0
MODULES=y
MODULE_UNLOAD=y
OBSOLETE_MODPARM=y
KMOD=y
STOP_MACHINE=y
X86_PC=y
M686=y
X86_CMPXCHG=y
X86_XADD=y
X86_L1_CACHE_SHIFT=5
RWSEM_XCHGADD_ALGORITHM=y
GENERIC_CALIBRATE_DELAY=y
X86_PPRO_FENCE=y
X86_WP_WORKS_OK=y
X86_INVLPG=y
X86_BSWAP=y
X86_POPAD_OK=y
X86_GOOD_APIC=y
X86_USE_PPRO_CHECKSUM=y
SMP=y
NR_CPUS=2
PREEMPT_NONE=y
PREEMPT_BKL=y
X86_LOCAL_APIC=y
X86_IO_APIC=y
X86_TSC=y
X86_MCE=y
X86_MSR=m
X86_CPUID=m
DCDBAS=m
NOHIGHMEM=y
SELECT_MEMORY_MODEL=y
FLATMEM_MANUAL=y
FLATMEM=y
FLAT_NODE_MEM_MAP=y
MTRR=y
IRQBALANCE=y
SECCOMP=y
HZ_250=y
HZ=250
PHYSICAL_START=0x100000
PM=y
ACPI=y
ACPI_AC=m
ACPI_BATTERY=m
ACPI_BUTTON=m
ACPI_VIDEO=m
ACPI_FAN=m
ACPI_PROCESSOR=m
ACPI_THERMAL=m
ACPI_ASUS=m
ACPI_IBM=m
ACPI_TOSHIBA=m
ACPI_BLACKLIST_YEAR=0
ACPI_EC=y
ACPI_POWER=y
ACPI_SYSTEM=y
X86_PM_TIMER=y
CPU_FREQ=y
CPU_FREQ_TABLE=y
CPU_FREQ_STAT=y
CPU_FREQ_DEFAULT_GOV_PERFORMANCE=y
CPU_FREQ_GOV_PERFORMANCE=y
CPU_FREQ_GOV_POWERSAVE=m
CPU_FREQ_GOV_USERSPACE=m
X86_ACPI_CPUFREQ=m
X86_SPEEDSTEP_ICH=m
X86_SPEEDSTEP_LIB=m
PCI=y
PCI_GOANY=y
PCI_BIOS=y
PCI_DIRECT=y
PCI_MMCONFIG=y
ISA_DMA_API=y
BINFMT_ELF=y
BINFMT_AOUT=m
BINFMT_MISC=m
NET=y
PACKET=m
UNIX=y
INET=y
IP_FIB_HASH=y
SYN_COOKIES=y
INET_DIAG=y
INET_TCP_DIAG=y
TCP_CONG_BIC=y
NETFILTER=y
IP_NF_CONNTRACK=m
IP_NF_FTP=m
IP_NF_IRC=m
IP_NF_QUEUE=m
IP_NF_IPTABLES=m
IP_NF_MATCH_LIMIT=m
IP_NF_MATCH_MAC=m
IP_NF_MATCH_PKTTYPE=m
IP_NF_MATCH_MARK=m
IP_NF_MATCH_MULTIPORT=m
IP_NF_MATCH_TOS=m
IP_NF_MATCH_ECN=m
IP_NF_MATCH_DSCP=m
IP_NF_MATCH_AH_ESP=m
IP_NF_MATCH_LENGTH=m
IP_NF_MATCH_TTL=m
IP_NF_MATCH_TCPMSS=m
IP_NF_MATCH_HELPER=m
IP_NF_MATCH_STATE=m
IP_NF_MATCH_CONNTRACK=m
IP_NF_MATCH_OWNER=m
IP_NF_FILTER=m
IP_NF_TARGET_REJECT=m
IP_NF_TARGET_LOG=m
IP_NF_TARGET_ULOG=m
IP_NF_TARGET_TCPMSS=m
IP_NF_NAT=m
IP_NF_NAT_NEEDED=y
IP_NF_TARGET_MASQUERADE=m
IP_NF_TARGET_REDIRECT=m
IP_NF_NAT_IRC=m
IP_NF_NAT_FTP=m
IP_NF_MANGLE=m
IP_NF_TARGET_TOS=m
IP_NF_TARGET_ECN=m
IP_NF_TARGET_DSCP=m
IP_NF_TARGET_MARK=m
IP_NF_ARPTABLES=m
IP_NF_ARPFILTER=m
VLAN_8021Q=m
IRDA=m
IRLAN=m
IRNET=m
IRCOMM=m
IRDA_ULTRA=y
IRDA_CACHE_LAST_LSAP=y
IRDA_FAST_RR=y
IRTTY_SIR=m
USB_IRDA=m
BT=m
BT_HCIUSB=m
BT_HCIUSB_SCO=y
BT_HCIUART=m
BT_HCIUART_H4=y
BT_HCIUART_BCSP=y
BT_HCIUART_BCSP_TXCRC=y
BT_HCIBCM203X=m
BT_HCIBFUSB=m
BT_HCIVHCI=m
IEEE80211=m
IEEE80211_CRYPT_WEP=m
IEEE80211_CRYPT_CCMP=m
IEEE80211_CRYPT_TKIP=m
STANDALONE=y
PREVENT_FIRMWARE_BUILD=y
FW_LOADER=m
CONNECTOR=m
PARPORT=m
PARPORT_PC=m
PARPORT_NOT_PC=y
BLK_DEV_FD=y
BLK_DEV_LOOP=m
BLK_DEV_RAM=y
BLK_DEV_RAM_COUNT=16
BLK_DEV_RAM_SIZE=4096
IOSCHED_NOOP=y
IOSCHED_AS=y
IOSCHED_DEADLINE=y
IOSCHED_CFQ=y
IDE=y
BLK_DEV_IDE=y
BLK_DEV_IDEDISK=y
BLK_DEV_IDECD=y
BLK_DEV_IDESCSI=y
IDE_GENERIC=y
BLK_DEV_IDEPCI=y
IDEPCI_SHARE_IRQ=y
BLK_DEV_IDEDMA_PCI=y
IDEDMA_PCI_AUTO=y
BLK_DEV_PIIX=y
BLK_DEV_IDEDMA=y
IDEDMA_AUTO=y
SCSI=y
SCSI_PROC_FS=y
BLK_DEV_SD=m
BLK_DEV_SR=m
BLK_DEV_SR_VENDOR=y
CHR_DEV_SG=m
SCSI_CONSTANTS=y
SCSI_LOGGING=y
SCSI_SPI_ATTRS=y
SCSI_SYM53C8XX_2=y
SCSI_SYM53C8XX_DMA_ADDRESSING_MODE=1
SCSI_SYM53C8XX_DEFAULT_TAGS=16
SCSI_SYM53C8XX_MAX_TAGS=64
SCSI_QLA2XXX=y
NETDEVICES=y
DUMMY=m
NET_ETHERNET=y
MII=m
NET_TULIP=y
TULIP=m
NET_PCI=y
E100=m
TIGON3=m
NET_RADIO=y
STRIP=m
IPW2100=m
IPW2100_MONITOR=y
IPW2200=m
AIRO=m
HERMES=m
PLX_HERMES=m
TMD_HERMES=m
NORTEL_HERMES=m
PCI_HERMES=m
ATMEL=m
PCI_ATMEL=m
PRISM54=m
HOSTAP=m
HOSTAP_FIRMWARE=y
HOSTAP_PLX=m
HOSTAP_PCI=m
NET_WIRELESS=y
PLIP=m
PPP=m
PPP_ASYNC=m
PPP_DEFLATE=m
PPP_BSDCOMP=m
SLIP=m
SLIP_COMPRESSED=y
INPUT=y
INPUT_MOUSEDEV=y
INPUT_MOUSEDEV_PSAUX=y
INPUT_MOUSEDEV_SCREEN_X=1024
INPUT_MOUSEDEV_SCREEN_Y=768
INPUT_KEYBOARD=y
KEYBOARD_ATKBD=y
INPUT_MOUSE=y
MOUSE_PS2=m
MOUSE_SERIAL=m
SERIO=y
SERIO_I8042=y
SERIO_SERPORT=y
SERIO_PCIPS2=m
SERIO_LIBPS2=y
VT=y
VT_CONSOLE=y
HW_CONSOLE=y
SERIAL_8250=y
SERIAL_8250_CONSOLE=y
SERIAL_8250_NR_UARTS=4
SERIAL_CORE=y
SERIAL_CORE_CONSOLE=y
UNIX98_PTYS=y
LEGACY_PTYS=y
LEGACY_PTY_COUNT=256
PRINTER=m
LP_CONSOLE=y
IPMI_HANDLER=m
IPMI_DEVICE_INTERFACE=m
IPMI_SI=m
IPMI_WATCHDOG=m
NVRAM=m
RTC=y
AGP=m
AGP_INTEL=m
DRM=m
DRM_RADEON=m
HPET=y
HPET_RTC_IRQ=y
HPET_MMAP=y
I2C=m
I2C_ALGOBIT=m
I2C_I801=m
I2C_I810=m
W1=m
W1_THERM=m
HWMON=m
VGA_CONSOLE=y
DUMMY_CONSOLE=y
SOUND=m
SND=m
SND_TIMER=m
SND_PCM=m
SND_HWDEP=m
SND_RAWMIDI=m
SND_SEQUENCER=m
SND_SEQ_DUMMY=m
SND_OSSEMUL=y
SND_MIXER_OSS=m
SND_PCM_OSS=m
SND_SEQUENCER_OSS=y
SND_RTCTIMER=m
SND_SEQ_RTCTIMER_DEFAULT=y
SND_GENERIC_DRIVER=y
SND_MPU401_UART=m
SND_DUMMY=m
SND_VIRMIDI=m
SND_MTPAV=m
SND_SERIAL_U16550=m
SND_MPU401=m
SND_AC97_CODEC=m
SND_AC97_BUS=m
SND_INTEL8X0=m
SND_USB_AUDIO=m
USB_ARCH_HAS_HCD=y
USB_ARCH_HAS_OHCI=y
USB=m
USB_DEVICEFS=y
USB_EHCI_HCD=m
USB_OHCI_HCD=m
USB_OHCI_LITTLE_ENDIAN=y
USB_UHCI_HCD=m
USB_ACM=m
USB_PRINTER=m
USB_STORAGE=m
USB_STORAGE_DATAFAB=y
USB_STORAGE_FREECOM=y
USB_STORAGE_ISD200=y
USB_STORAGE_DPCM=y
USB_STORAGE_SDDR09=y
USB_STORAGE_SDDR55=y
USB_STORAGE_JUMPSHOT=y
USB_HID=m
USB_HIDINPUT=y
HID_FF=y
USB_HIDDEV=y
USB_KBD=m
USB_MOUSE=m
USB_AIPTEK=m
USB_WACOM=m
USB_KBTAB=m
USB_POWERMATE=m
USB_MTOUCH=m
USB_EGALAX=m
USB_XPAD=m
USB_ATI_REMOTE=m
USB_MDC800=m
USB_MICROTEK=m
USB_DABUSB=m
USB_CATC=m
USB_KAWETH=m
USB_PEGASUS=m
USB_RTL8150=m
USB_USBNET=m
USB_NET_AX8817X=m
USB_NET_CDCETHER=m
USB_NET_NET1080=m
USB_NET_ZAURUS=m
USB_MON=y
USB_USS720=m
USB_SERIAL=m
USB_SERIAL_GENERIC=y
USB_SERIAL_BELKIN=m
USB_SERIAL_DIGI_ACCELEPORT=m
USB_SERIAL_EMPEG=m
USB_SERIAL_FTDI_SIO=m
USB_SERIAL_VISOR=m
USB_SERIAL_IPAQ=m
USB_SERIAL_IR=m
USB_SERIAL_EDGEPORT=m
USB_SERIAL_EDGEPORT_TI=m
USB_SERIAL_KEYSPAN_PDA=m
USB_SERIAL_KEYSPAN=m
USB_SERIAL_KEYSPAN_MPR=y
USB_SERIAL_KEYSPAN_USA28=y
USB_SERIAL_KEYSPAN_USA28X=y
USB_SERIAL_KEYSPAN_USA28XA=y
USB_SERIAL_KEYSPAN_USA28XB=y
USB_SERIAL_KEYSPAN_USA19=y
USB_SERIAL_KEYSPAN_USA18X=y
USB_SERIAL_KEYSPAN_USA19W=y
USB_SERIAL_KEYSPAN_USA19QW=y
USB_SERIAL_KEYSPAN_USA19QI=y
USB_SERIAL_KEYSPAN_USA49W=y
USB_SERIAL_KEYSPAN_USA49WLC=y
USB_SERIAL_KLSI=m
USB_SERIAL_KOBIL_SCT=m
USB_SERIAL_MCT_U232=m
USB_SERIAL_PL2303=m
USB_SERIAL_SAFE=m
USB_SERIAL_SAFE_PADDED=y
USB_SERIAL_CYBERJACK=m
USB_SERIAL_XIRCOM=m
USB_SERIAL_OMNINET=m
USB_EZUSB=y
USB_EMI62=m
USB_EMI26=m
USB_AUERSWALD=m
USB_RIO500=m
USB_LEGOTOWER=m
USB_LCD=m
USB_LED=m
USB_CYTHERM=m
USB_PHIDGETSERVO=m
USB_TEST=m
EXT2_FS=y
FS_POSIX_ACL=y
XFS_FS=y
XFS_EXPORT=y
XFS_QUOTA=y
MINIX_FS=m
ROMFS_FS=m
INOTIFY=y
QUOTACTL=y
DNOTIFY=y
AUTOFS_FS=m
AUTOFS4_FS=m
FUSE_FS=m
ISO9660_FS=m
JOLIET=y
ZISOFS=y
ZISOFS_FS=m
UDF_FS=m
UDF_NLS=y
FAT_FS=m
MSDOS_FS=m
VFAT_FS=m
FAT_DEFAULT_CODEPAGE=437
FAT_DEFAULT_IOCHARSET="iso8859-1"
NTFS_FS=m
PROC_FS=y
PROC_KCORE=y
SYSFS=y
TMPFS=y
RAMFS=y
RELAYFS_FS=m
EFS_FS=m
HPFS_FS=m
UFS_FS=m
NFS_FS=m
NFS_V3=y
NFS_V4=y
NFSD=m
NFSD_V3=y
NFSD_V4=y
NFSD_TCP=y
LOCKD=m
LOCKD_V4=y
EXPORTFS=y
NFS_COMMON=y
SUNRPC=m
SUNRPC_GSS=m
RPCSEC_GSS_KRB5=m
SMB_FS=m
MSDOS_PARTITION=y
NLS=y
NLS_DEFAULT="iso8859-1"
NLS_CODEPAGE_437=m
NLS_ISO8859_1=m
NLS_UTF8=m
DEBUG_KERNEL=y
MAGIC_SYSRQ=y
LOG_BUF_SHIFT=16
DETECT_SOFTLOCKUP=y
DEBUG_BUGVERBOSE=y
EARLY_PRINTK=y
X86_FIND_SMP_CONFIG=y
X86_MPPARSE=y
KDB=y
KDB_MODULES=y
KDB_CONTINUE_CATASTROPHIC=0
CRYPTO=y
CRYPTO_MD5=y
CRYPTO_DES=m
CRYPTO_AES=m
CRYPTO_ARC4=m
CRYPTO_MICHAEL_MIC=m
CRC_CCITT=m
CRC32=m
ZLIB_INFLATE=m
ZLIB_DEFLATE=m
GENERIC_HARDIRQS=y
GENERIC_IRQ_PROBE=y
GENERIC_PENDING_IRQ=y
X86_SMP=y
X86_HT=y
X86_BIOS_REBOOT=y
X86_TRAMPOLINE=y
PC=y

#1 lore raw
Quoted 2 lines · Keith Owens

fs/bio.c: In function 'bio_alloc_bioset':
fs/bio.c:167: warning: 'idx' may be used uninitialized in this function

This is bogus. bio_alloc_bioset() passes in a pointer to a long which bvec_alloc_bs() always fills, _unless_ it returns NULL. In the NULL case we never use 'idx' (quite obvious, we bail immediately).

I didn't check the other warnings, but there may be similar cases in other paths.

Signature · 1 line
Jens Axboe
#2 lore raw
In reply to Keith Owens
Quoted 7 lines · Keith Owens

gcc 4.0.1 on i386, 2.6.14 assorted warnings.

arch/i386/kernel/cpu/transmeta.c: In function 'init_transmeta':
arch/i386/kernel/cpu/transmeta.c:11: warning: 'cpu_freq' may be used uninitialized in this function

fs/bio.c: In function 'bio_alloc_bioset':
fs/bio.c:167: warning: 'idx' may be used uninitialized in this function

gcc is dumb, it doesn't realise that the variable will be filled by another function if its passed thus..

	unsigned long foo
	bar(&foo)
	if (foo==1)
		...

With bar() filling in content of foo.
I believe there's at least once instance of this in gcc bugzilla.

Dave

#3 lore raw
Quoted 12 lines · Dave Jones

gcc is dumb, it doesn't realise that the variable will be filled by another function if its passed thus..

	unsigned long foo
	bar(&foo)
	if (foo==1)
		...

With bar() filling in content of foo.
I believe there's at least once instance of this in gcc bugzilla.

Can we blame the compiler not knowing if the function bar will use the value of foo or initialize it? For these cases, would it not be better to set them a value (for example 0 or maybe UNINIT_DEFAULT)? After all, the most common warning (as far as i can see) is about uninitialized variables.

Richard

#4 lore raw
In reply to Dave Jones
Quoted 11 lines · Dave Jones

gcc is dumb, it doesn't realise that the variable will be filled by another function if its passed thus..

        unsigned long foo
        bar(&foo)
        if (foo==1)
                ...

With bar() filling in content of foo.
I believe there's at least once instance of this in gcc bugzilla.

The case is more complicated here. gcc before 4.0 always assumed that bar(&foo); initializes foo. gcc-4.0 now looks into bar if it is inlining that and tries to find out if that really initializes foo in every case.

In the example, bvec_alloc_bs does not initialize &idx when nr is not between 1 and BIO_MAX_PAGES, so gcc is telling the truth here.

	Arnd <><
#5 lore raw
Quoted 2 lines · Arnd Bergmann

In the example, bvec_alloc_bs does not initialize &idx when nr is not between 1 and BIO_MAX_PAGES, so gcc is telling the truth here.

Wrong. idx is always initialized if being used.

Signature · 1 line
Jens Axboe
#6 lore raw
Quoted 12 lines · Jens Axboe
Quoted 4 lines · Arnd Bergmann

In the example, bvec_alloc_bs does not initialize &idx when nr is not between 1 and BIO_MAX_PAGES, so gcc is telling the truth here.

Wrong. idx is always initialized if being used.

Is the compiler really that smart as it searches back into the parent-function and try all the combinations? Otherwise, Arnd is correct. And on an philosophical plane, can/should we put that responsibility onto the compiler? Is it not "easier" to make the functions take care of its own duties (like the *nix-way) and make the bvec_alloc_bs initialize idx (even if it has to be an error-value)?

I'm thinking something like this. Seems alright?

/Richard

---

fs/bio.c +0 -0
diff -Nurp a/fs/bio.c b/fs/bio.c
@@ -90,6 +90,7 @@ static inline struct bio_vec *bvec_alloc
		case  65 ... 128: *idx = 4; break;
		case 129 ... BIO_MAX_PAGES: *idx = 5; break;
		default:
+			*idx = -1;
			return NULL;
	}
	/* 
#7 lore raw
Quoted 6 lines · Richard Knutsson

And on an philosophical plane, can/should we put that responsibility onto the compiler? Is it not "easier" to make the functions take care of its own duties (like the *nix-way) and make the bvec_alloc_bs initialize idx (even if it has to be an error-value)?

I'm thinking something like this. Seems alright?

No. Working around the false positives in compiler warning system is *wrong*. _IF_ it cares to inline the function and generate the warnings based on that, it is responsible for doing it right.

It's a gcc bug, plain and simple.

#8 lore raw
In reply to Dave Jones
Quoted 10 lines · Dave Jones

gcc is dumb, it doesn't realise that the variable will be filled by another function if its passed thus..

	unsigned long foo
	bar(&foo)
	if (foo==1)
		...

With bar() filling in content of foo.
I believe there's at least once instance of this in gcc bugzilla.

gcc is a *LOT* smarter than you give it credit for. It will not warn for cases where it isn't able to tell how foo is used passed with &foo. It will warn for cases where it can

Compile the following program with and without -O2 and you'll see just how smart gcc actually is

#include <stdio.h>

static void bar(char *x) {
*x = 1;
}

static void baz(char *x) {
printf("Wombat\n");
}

int main(int argc, char *argv[]) {
char x,y;
bar(&x);
baz(&y);

printf("%c,%c\n", x, y);
}

So I'd go look deeper for real bugs

Alan

#9 lore raw
Quoted 3 lines · Alan Cox

gcc is a *LOT* smarter than you give it credit for. It will not warn for cases where it isn't able to tell how foo is used passed with &foo. It will warn for cases where it can

static inline __attribute__((always_inline)) int bar(int n, int *idx)

{
        if (n != 1) 
                return 0; 
        *idx = 1;
        return 1;  
}
void baz(int *v, int n)
{
        int idx;
        int p = bar(n, &idx);
        if (__builtin_expect(!p, 0))
                return; 
        *v |= idx;
}

and try gcc -Wall on that. Watch it warn about idx being possibly uninitialized. Replace __builtin_expect(!p, 0) with !p and see the warning go away. That, BTW, is a trimmed-down source of bogus warning in bio.c. And a lot of its analogs all over the tree.

#10 lore raw
Quoted 16 lines · Al Viro

static inline __attribute__((always_inline)) int bar(int n, int *idx)

{
        if (n != 1) 
                return 0; 
        *idx = 1;
        return 1;  
}
void baz(int *v, int n)
{
        int idx;
        int p = bar(n, &idx);
        if (__builtin_expect(!p, 0))
                return; 
        *v |= idx;
}
Quoted 1 line

and try gcc -Wall on that.

Duh. gcc -Wall -Os, of course

#11 lore raw

Also, this test case:

------------------------------------------------------------
extern int f1(void);

static unsigned char f2(unsigned int *def)

{
	if (!f1())
		return 0;
	*def = 1;
	return 1;
}

unsigned char f3(unsigned char **eoc)

{
	unsigned int def;

	if (!f2(&def))
		return 0;
	if (def)
		*eoc = (unsigned char *)1;
	else
		*eoc = 0;
	return 1;
}

static int f4(unsigned int *def)

{
	if (!f1())
		return 0;
	*def = 1;
	return 1;
}

unsigned char f5(unsigned char **eoc)

{
	unsigned int def;

	if (!f4(&def))
		return 0;
	if (def)
		*eoc = (unsigned char *)1;
	else
		*eoc = 0;
	return 1;
}

------------------------------------------------------------

compiled with "gcc -Wall -O2" gives:
xx.c: In function 'f3':
xx.c:14: warning: 'def' may be used uninitialized in this function

but doesn't complain about the same code in f5. The difference is that f4 returns "int" while f2 returns "unsigned char". This has apparently been fixed in gcc mainline - the above comes from Debian unstables' gcc (4.0.3 20051023 (prerelease) (Debian 4.0.2-3)).

Compiling with "gcc -Wall -O2 -fnoinline" suppresses the warning.

Signature · 3 lines
Cheers,
Stephen Rothwell                    [email protected]
http://www.canb.auug.org.au/~sfr/
#12 lore raw
In reply to Alan Cox
Quoted 3 lines · Alan Cox

gcc is a *LOT* smarter than you give it credit for. It will not warn for cases where it isn't able to tell how foo is used passed with &foo. It will warn for cases where it can

Gcc thinks it is a lot smarter than it actually is, in fact. :)

The usual case where it gives a bogus warning is where the value of one variable is correlated with the defined-ness of another, for example:

	found = 0;
	for (i = 0; i < N; ++i) {
		if (a[i] == foo) {
			x = b[i];
			found = 1;
			break;
		}
	}
	if (found)
		bar(x);

Gcc will warn on the use of x when in fact it is perfectly OK, and we get quite a few of these in compiling a kernel. At a minimum, I would like to be able to disable the "may be used uninitialized" warnings while still getting the "is used uninitialized" warnings.

Paul.

#13 lore raw
Quoted 4 lines · Paul Mackerras

Gcc will warn on the use of x when in fact it is perfectly OK, and we get quite a few of these in compiling a kernel. At a minimum, I would like to be able to disable the "may be used uninitialized" warnings while still getting the "is used uninitialized" warnings.

Quite. "may be used uninitialized" has _way_ too low S/N ratio - absolute majority of these warnings are bogus. Moreover, several very common use patterns reliably trigger them.