[bug] dd piped input truncates ~1% of values to %4096 on multi-core systems

[John Thomson]

Hi Busybox,

I may have run into a bug where dd with piped input will truncate a small number of results (~1%) to a modulo 4096 value on multi-core systems.
The test script does not produce errors running under bash.
This issue is not seen if dd reads the file with if=
I have not looked into the cause.

Example:
# generate test file
dd if=/dev/urandom of=/tmp/wlan_data_0 bs=1M count=1
# ~1% pipe error
cat /tmp/wlan_data_0 | \
dd iflag=skip_bytes bs=$((0x2f20)) skip=0 count=1 2>/dev/null | wc -c

First noticed on arm with busybox 1.31.1
Reproduced on x86_64 busybox g6a55b4e40
Reproduced on mipsel 4 core
Not reproduced on mips single core

cat data | dd used for Openwrt wifi radio calibration data extraction from sysfs:
https://github.com/openwrt/openwrt/commit/7557e7f267e845db5a403139c49f2637f9021992

Scripted example:
test_caldata_extract() {
	local counter
	local count_good
	local count_bad
	local result
	local result_latest
	local dd_bs

	dd_bs=$((0x2f20))

	dd if=/dev/urandom of=/tmp/wlan_data_0 bs=1M count=1
	result=$(cat /tmp/wlan_data_0 | \
	dd iflag=skip_bytes bs=$dd_bs skip=0 count=1 2>/dev/null | wc -c)
	while [ ! $result -eq $dd_bs ]; do
	echo "bad initial result"
	result=$(cat /tmp/wlan_data_0 | \
	dd iflag=skip_bytes bs=$dd_bs skip=0 count=1 2>/dev/null | wc -c)
	done

	counter=0
	count_good=0
	count_bad=0
	echo "initial run $counter $result"

	while true; do
	counter=$(( counter + 1 ))
	result_latest=$(cat /tmp/wlan_data_0 | \
	dd iflag=skip_bytes bs=$((0x2f20)) skip=0 count=1 2>/dev/null | wc -c)
	if [ "$result_latest" = "$result" ]; then
	count_good=$(( count_good + 1 ))
	#echo "run $counter okay"
	else
	count_bad=$(( count_bad + 1 ))
	echo "run $counter bad $result_latest ($count_bad / $counter)"
	#break
	fi
	done
}

test_caldata_extract


Cheers
-- 
  John Thomson