Skip to contentSkip to conversation

Multiple & vs. && and | vs. || bugs in 2.4 and 2.5

DISCUSSION Norbert Kiesel 4 participants 4 messages Mar 2, 2003

lore
Share & tools
lore
#0 lore raw

Hi,
inspired by the recent bug report from Muli Ben-Yehuda (see
http://marc.theaimsgroup.com/?l=linux-kernel&m=104647477025930), I hunted for similar bugs in 2.4 and 2.5 using a simple grep:

find ${1:-.} -name \*.c | xargs grep -En '![a-zA-Z0-9_ ]+(\|[^|]|\&[^&])|([^|]\||[^&]\&) *!'

This found typos in both 2.4.20 and 2.5.63. Please note that is a trimmed down list of the grep, I removed what I considered false positives (e.g. real bit operations and matches within comments). All but the trident.c typo looks like a logical and not a bit operation to me.

linux-2.4.20/drivers/net/aironet4500_core.c:2679: if (adhoc & !max_mtu)
linux-2.4.20/drivers/sound/gus_midi.c:186: return (qlen > 0) | !(GUS_MIDI_STATUS() & MIDI_XMIT_EMPTY);
linux-2.4.20/drivers/sound/gus_wave.c:3126: if ((gus_mem_size > 0) & !gus_no_wave_dma)
linux-2.4.20/drivers/sound/maestro.c:3362: if(! w & PCI_STATUS_CAP_LIST) return 0;
linux-2.4.20/drivers/sound/trident.c:3063: if(!wcontrol & 0x8000)
linux-2.4.20/drivers/video/aty128fb.c:2534: if (!reg & LVDS_ON) {
linux-2.4.20/drivers/video/radeonfb.c:2781: if (!lvds_gen_cntl & LVDS_ON) {
linux-2.4.20/drivers/usb/acm.c:243: if (!urb->status & !acm->throttle) {

linux-2.4.20/drivers/i2c/i2c-proc.c:732:			       SENSORS_ANY_I2C_BUS) & !is_isa))
linux-2.4.20/drivers/i2c/i2c-proc.c:821:			       probe[i] == SENSORS_ANY_I2C_BUS) & !is_isa))
linux-2.4.20/drivers/i2c/i2c-proc.c:838:			       SENSORS_ANY_I2C_BUS) & !is_isa))

linux-2.5.63/drivers/media/dvb/dvb-core/dvb_demux.c:224: if (f->doneq & !neq)
linux-2.5.63/drivers/video/console/fbcon.c:459: if (rw & !bottom_only) {
linux-2.5.63/drivers/serial/uart00.c:238: if (!status & (UART_MSR_DCTS_MSK | UART_MSR_DDSR_MSK |
linux-2.5.63/drivers/pnp/pnpbios/core.c:325: if ( !boot & pnpbios_dont_use_current_config )
linux-2.5.63/drivers/pnp/pnpbios/core.c:353: if ( !boot & pnpbios_dont_use_current_config )
linux-2.5.63/net/sctp/tsnmap.c:249: if (started & !ended) {
linux-2.5.63/sound/oss/gus_midi.c:182: return (qlen > 0) | !(GUS_MIDI_STATUS() & MIDI_XMIT_EMPTY);
linux-2.5.63/mm/slab.c:1646: if (!flags & __GFP_WAIT)

Should I create patches for these and send them to the maintainers?

so long
Norbert

#1 lore raw
Quoted 1 line · Norbert Kiesel

Should I create patches for these and send them to the maintainers?

Yes.

-- wli

#2 lore raw
In reply to Norbert Kiesel
Quoted 4 lines · Norbert Kiesel

Hi,
inspired by the recent bug report from Muli Ben-Yehuda (see
http://marc.theaimsgroup.com/?l=linux-kernel&m=104647477025930), I hunted for similar bugs in 2.4 and 2.5 using a simple grep:

In the interest of giving credit where credit is due, John Levon found the bug and sent me the patch. He did this for several other drivers as well.

Quoted 1 line

Should I create patches for these and send them to the maintainers?

Definitely. You can ignore the 2.4.20 trident one, I'll fix it and push it to Marcelo.

Thanks,
Muli.

Signature · 2 lines
Muli Ben-Yehuda
http://www.mulix.org
#3 lore raw
In reply to Norbert Kiesel
Quoted 2 lines · Norbert Kiesel

linux-2.4.20/drivers/video/aty128fb.c:2534: if (!reg & LVDS_ON) {
linux-2.4.20/drivers/video/radeonfb.c:2781: if (!lvds_gen_cntl & LVDS_ON) {

There two are single bit tests, they should probably be written with an additional parenthesis:

	if (!(reg & LVDS_ON))
	if (!(lvds_gen_cntl & LVDS_ON))

Though in these specific cases, I tend to prefer writing it this way

	if ((reg & LVDS_ON) == 0)
	if ((lvds_gen_cntrl & LVDS_ON) == 0)

Since I'm the author of both of the above (though I'm not the driver maintainer), feel free to send the typo fixes to Marcelo along with Ani Joshi (driver maintainer)

Ben.