Run LTP
===
= Git Repo =
atclnx01:/home/project/git_repo/os/ltp.git
v3 branch:
20130109-nds32
v5 branch:
20180316-riscv

= Quick start =
Accessing LTP folder by NFS and the /tmp folder is on SD card.

One the Linux kernel booting is finished, typing following commands.
# date -s YYYYMMDDHHMM (date -s 201612311830)
#: (note: setting the date )
# ifconfig eth0 hw ether 00:23:96:00:XX:XX
#: (note: setting the MAC address for board. XX:XX means user-defined variable(using Staff number is better) )
# udhcpc
# mount -t nfs -o nolock,rsize=1024,wsize=1024 <LTP server IP>:<LTP folder path on sever> <folder on board>
#: (note:mount LTP. ex: mount -t nfs -o nolock,rsize=1024,wsize=1024 atcsqa05:/sqa5a/ltp/BSPv321/2014-04-10/nds32le-linux-glibc-v3-O0 /mnt )
# fdisk -l
#: (note: check SD driver version. if nothing to show, go following step. 1. otherwise, go step 2.)
## mount /dev/cf /tmp (SD 1.0)
## mount /dev/mmcblk0p1 /tmp (SD 2.0)
##: (note: mount /tmp on SD card)
# cd /mnt/Deploy/opt/ltp
# ./runltp -q 2>&1 |tee ./results/runXXXXX.log
#: (run ltp and the log is recorded in ./results/runXXXXX.log)

= LTP Execution =
* Checkout from LTP git os repository and latest version branch!
* Build the LTP program by using "sh build_ltp.sh"
* Execute using the runltp script in LTP directory.
** Full execution: ./runltp -q
** Execute a specific testcase (-s): ./runltp -q -s access01
** Execute specific category (-f): ./runltp -q -f syscalls
*** Categories are

${LTPROOT}/runtest/syscalls \
${LTPROOT}/runtest/fs \
${LTPROOT}/runtest/fsx \
${LTPROOT}/runtest/dio \
${LTPROOT}/runtest/io \
${LTPROOT}/runtest/mm \
${LTPROOT}/runtest/ipc \
${LTPROOT}/runtest/sched \
${LTPROOT}/runtest/math \
${LTPROOT}/runtest/nptl \
${LTPROOT}/runtest/pty \
${LTPROOT}/runtest/containers \
${LTPROOT}/runtest/fs_bind \
${LTPROOT}/runtest/controllers \
${LTPROOT}/runtest/filecaps \
${LTPROOT}/runtest/cap_bounds \
${LTPROOT}/runtest/fcntl-locktests \
${LTPROOT}/runtest/connectors \
${LTPROOT}/runtest/admin_tools \
${LTPROOT}/runtest/timers \
${LTPROOT}/runtest/power_management_tests \
${LTPROOT}/runtest/numa \
${LTPROOT}/runtest/hugetlb \
${LTPROOT}/runtest/commands \
${LTPROOT}/runtest/hyperthreading

* Execute a specific testcase by config file : Deploy/opt/ltp/scenario_groups/default
syscalls
#fs // ignore this case by prefix #
fs_perms_simple
...

* Brief introduction : T:\RD-SW\SI\LTP\LTP_brief_introduction.ppt

Cautions:
1. Don't use the SD card whose file system is vfat for tmp directory. chmod does not work on it
2. The mount order:
step 1. mount ltp folder
step 2. mount /tmp folder

= Format SD card to ext2 =
User can follow the below steps to partition SD and format it to ext2 on server.
# Find targeted SD card by fdisk
#* command: sudo fdisk -l
# Low-level format SD card (optional)
#* command: sudo dd if=/dev/zero of=/dev/<target device>
#*: ex: sudo dd if=/dev/zero of=/dev/sdc
# Create a new partition table on SD (skip this step if low-level format SD card is not executed)
#*command:
#*# sudo fdisk /dev/<target device>
#*# m (list all command)
#*# o (create a new empty DOS partition table)
#*# w
# Add a new partition to SD card
#*command:
#*# sudo fdisk /dev/<target device>
#*#: ex: sudo fdisk /dev/sdc
#*# n (add a new partition)
#*# p
#*# 1 (or 2,3,4...)
#*# <enter> (define the start address of this partition. In this case, default value is used)
#*# +1G (define the size of this partition. In this case, the partition size is 2GB)
#*# t (specify the partition system ID from previous step's result)
#*# L (list all partition system ID)
#*# 83 (Linux)
#*# Repeat step 2. ~ step 9. to create needed partition.
#*# p (list all partition table on SD card.)(optional)
#*# w (write to SD card)
# format targeted partition to ext2 on SD card
#*command:
#*# sudo fdisk -l /dev/<target device>
#*#: list all partition on target device
#*# sudo mkfs -t ext2 /dev/<target device><partition number>
#*#: ex: sudo mkfs -t ext2 /dev/sdc1

= LTP Update Policy =
* We will pull LTP official git repository to our git server periodically.
* Each official LTP release will be tagged and created a branch available for testing.
** SQA can selectively choose a version to test.
* LTP's bug will be patched in the branch.

= LTP Reports =
* 20190501 [[LTP/20190501]]
* 20190408 Pixelplus: [[LTP/20190408]]
* Goldens: T:\RD-SW\SI\LTP\golden
** 20190410 A report of all main stream setup: [[LTP/20190410]]
* Legacy reports: T:\RD-SW\SI\LTP

= LTP False Positives =
== 20180824 ==
#Environment:
##Kernel: RISCV-Linux-4.17 (e7f0980ad6ed423ddb298810f8a776a5f5e993b9)
##Toolchain: /NOBACKUP/atcsqa06/alankao/nds64le-linux-glibc-v5d-180827
##repo: atclnx01:/home/project/git_repo/os/ltp.git
##branch: 20180316-riscv
##LTP located in SD Card
### assumed to be in /dev/mmcblk0, mounted at /mnt
#Setup: using ae350_64_defconfig
#False positive case
## timing-related cases such as follows. overhead can not be covered within given threshold.
### clock_nanosleep02
### epoll_wait02
### nanosleep01
### poll02
### pselect01/02
### select04
### futex_wait05
### futex_wait_bitset01/02
## stress test on cgroup cpu subsystem. Lack of synchronization between testing processes causes main process try to kill an unborn program. To make these tests work, a timeout is added after launching program in cgroup_fj_stress.sh to make sure all programs are ready to be killed.
### cgroup_fj_stress_cpu_2_2_each
### cgroup_fj_stress_cpu_3_3_each
### cgroup_fj_stress_cpu_4_4_each
### cgroup_fj_stress_cpu_2_9_each
### cgroup_fj_stress_cpu_10_3_each
### cgroup_fj_stress_cpu_1_200_each
### cgroup_fj_stress_cpu_200_1_each
## oom01
##* This testcase possibly terminates whole LTP testing. oom01 suppose to test the OOM handling flow so it means to generate out-of-memory cases. Because OOM is a critical event for kernel, /sbin/syslogd will got the information from /sbin/klogd and record it on a particular file if OOM is happened. Unfortunately, another OOM event is possibly triggered by /sbin/syslogd if the released memory by OOM reaper is not adequate for /sbin/syslogd. In this condition, OOM killer will select a process to kill again and ./runltp is always the sacrifice. Hence LTP is terminated.

== 20180316-riscv ==
#Environment:
##Kernel: 4.17 (3182b589f97b8cf2be646f941ab8e182af6db490)
##Toolchain: BSP-latest nds64le-linux-glibc-v5 date:2018/4/9
##repo: atclnx01:/home/project/git_repo/os/ltp.git
##branch: 20180316-riscv
##LTP can be located in SD Card
### assumed to be in /dev/mmcblk0p1, mounted at /mnt
#Setup: using ae350_64_defconfig
#Ignore Cases (Suggested)
## Too long
### syscalls/fork13
### syscalls/msgctl09~11
### mm/mallocstress
### mm/mmapstress09
### math/float_*
### nptl/nptl01

#False positive case
#* getrandom01
#** When passing GRAN_RANDOM to this syscall, it fetches random data from /dev/random instead of /dev/urandom, and blocks due to the blocking nature of /dev/random. Since the environment is often clean, no enough entropy would be generated on time, so it almost always fails because of timeout.
#* keyctl02
#** It can pass if we double the TIMEOUT.

#Strange case
#* ptem01, hangup01
#** Although they said that they need to find /dev/ptmx, mounting /dev/pts as a devpts file system is the key to make it pass.
#* timer-related test like leapsec_timer or clock_settime02
#** sometimes PASS, sometimes FAIL

== 20170427 ==
#Environment:
##Kernel: 3.10(9058ce2867bbbe68e6205d34eb133319f3afc64f)
##Toolchain: BSP420 glibc v3 toolchain.
##/tmp is mount on SD card
##LTP is located on atcsqa02 and accessed by NFS
#Setup:
#:The following kernel configure shall be enabled before running testing.
## CONFIG_CGROUPS
## CONFIG_CGROUP_CPUACCT
## CONFIG_IPV6
## CONFIG_FANOTIFY
## CONFIG_MINIX_FS
## CONFIG_VETH
## CONFIG_FILE_LOCKING
#Long time testing:
#: syscalls, mm, sched, fs, math
#False positive case:
#:The false positive case are listed as below. For new false positive cases in LTP 2017, the reason will be appended to this list.
#* remap_file_pages01 FAIL 7
#** A false positive case in aliasing cache CPU, because arch_get_unmapped_area() will check "if (aliasing && flags & MAP_SHARED && (addr - (pgoff << PAGE_SHIFT)) & (SHMLBA -1)"
#** Need to setup environment before testing
#*** mkdir -p /dev/shm
#*** mount -t tmpfs none /dev/shm
#*** /mnt/greentime/contrib/ltp/Deploy/opt/ltp # df -h
#*** Filesystem Size Used Available Use% Mounted on
#*** atcsqa02:/sqa2 2.6T 1.3T 1.2T 52% /mnt
#*** none 504.7M 0 504.7M 0% /dev/shm
#** It shall pass if it is not run on a alising CPU.
#* getrusage04 FAIL 1
#* fanotify06 FAIL 7
#** [bug 13749], The targeted kernel commit for this testcase is not included in kernel 3.10
#* move_pages03 FAIL 1
#* move_pages11 FAIL 1
#* prot_hsymlinks FAIL 11
#** Busybox does not support utility useradd,userdel but testcase needs them.
#* pselect01 FAIL 1
#* pselect01_64 FAIL 1
#** The error message of above two testcases are "slept too long". We think the timing precision issue is due that the lower operating frequency on FPGA borad.
#* recvmsg02 FAIL 2
#** [bug 13738]. The fixed commit is not included in kernel 3.10
#* setpriority01 FAIL 6
#** Busybox does not support utility useradd,userdel but testcase needs them.
#* syslog01 FAIL 1
#* syslog02 FAIL 1
#* syslog03 FAIL 1
#* syslog04 FAIL 1
#* syslog05 FAIL 1
#* syslog06 FAIL 1
#* syslog07 FAIL 1
#* syslog08 FAIL 1
#* syslog09 FAIL 1
#* syslog10 FAIL 1
#* utimensat01 FAIL 2
#* isofs FAIL 1
#** mkfs.iso is needed by testcase but busybox does not support
#* aio01 FAIL 2
#**glibc does not support libaio
#* aio02 FAIL 2
#**glibc does not support libaio
#* netns_breakns_ip_ipv4_netlink FAIL 2
#* netns_breakns_ip_ipv6_netlink FAIL 2
#* netns_breakns_ip_ipv4_ioctl FAIL 2
#* netns_breakns_ip_ipv6_ioctl FAIL 2
#* netns_comm_ip_ipv4_netlink FAIL 2
#* netns_comm_ip_ipv6_netlink FAIL 2
#* netns_comm_ip_ipv4_ioctl FAIL 2
#* netns_comm_ip_ipv6_ioctl FAIL 2
#* netns_sysfs FAIL 2
#** The above 9 testcases, netns_XXX.. , are failed due that the option "-V" is not supported by utility ip in busybox
#* su01 FAIL 1
#* cron02 FAIL 1
#* cron_deny01 FAIL 1
#* cron_allow01 FAIL 1
#* cron_dirs_checks01 FAIL 1
#* acl_test01 FAIL 2
#* move_pages03 FAIL 1
#* move_pages11 FAIL 1
#* ar FAIL 3
#* file01 FAIL 2
#* tar01 FAIL 3
#* cron FAIL 2
#* logrotate FAIL 1
#* cpio01 FAIL 1
#* unzip01 FAIL 1
#* gzip01 FAIL 1
#* mv01 FAIL 2
#* df01_ext3 FAIL 6
#** mkfs.ext3 is unsupported by busybox
#* df01_ext4 FAIL 6
#** mkfs.ext4 is unsupported by busybox
#* df01_xfs FAIL 6
#** mkfs.xfs is unsupported by busybox
#* df01_exfat FAIL 6
#** mkfs.exfat is unsupported by busybox
#* df01_ntfs FAIL 6
#** mkfs.ntfs is unsupported by busybox
#* mkfs01_minix FAIL 2
#** mkfs.minix is unsupported by busybox
#* which01 FAIL 33
#*# In busybox's "utility which", "utility which" does not expend "." to current folder path in output
#*# option: --all, --read-alias, -i, --version, -v, -V is not supported by busybox's "utility which".
#* lsmod01 FAIL 1
#** If no module has been inserted before running testcase, testcase consider that lsmod does not work and #report fail.
#* wc01 FAIL 1
#** Option --bytes,--lines,--max-line-length,--words,--chars,--help,--version are not supported by wc in busybox
#* getaddrinfo_01 FAIL 1
#** Tester shall add "<IPV4 address> Andes" and "<IPV6 address> Andes" to /etc/hosts due that The DNS server does not have ip address for "Andes".
# CONF cases
#:522 CONF cases which LTP thinks these cases are failed due that relative kernel configurations are disable. It is pending now.

== 20130109 ==
#Part1: test with kernel 3.4.35(9058ce2867bbbe68e6205d34eb133319f3afc64f) and BSP410 toolchain
#* getrusage04 FAIL 1
#* profil01 FAIL 1
#** Bug 13927. This testcase is only for dynamic linking program. In addition, this testcase needs following two modifications to make testing work.
#**# ADDRESS_OFFSET shall be set to 0x8000 for nds32.
#**# The command to disable profil shall be profil(NULL,,,) not profil(,,,0).
#* readahead02 FAIL 1
#** This testcase should be failed on tmpfs. It can pass if you don't mount tmp to tmpfs or you mount tmp to SD card.
#* sysctl03 FAIL 4
#* syslog01 FAIL 1
#* syslog02 FAIL 1
#* syslog03 FAIL 1
#* syslog04 FAIL 1
#* syslog05 FAIL 1
#* syslog06 FAIL 1
#* syslog07 FAIL 1
#* syslog08 FAIL 1
#* syslog09 FAIL 1
#* syslog10 FAIL 1
#* utimensat01 FAIL 1
#* mtest06_2 FAIL 255
#* thp02 FAIL 2
#* max_map_count FAIL 1
#* su01 FAIL 1
#* cron02 FAIL 1
#* cron_deny01 FAIL 1
#* cron_allow01 FAIL 1
#* cron_dirs_checks01 FAIL 1
#* acl_test01 FAIL 1
#* Numa-testcases FAIL 2 or 1
#** The returned error number is 2 if /bin/sh is redirected to /bin/busybox
#**: (The error is due to numa01.sh: line 438: syntax error: bad substitution )
#** The returned error number is 1 if /bin/sh is redirected to /bin/bash.
#**: (The error is due that "numactl" is not found)
#* hugemmap01 FAIL 2
#* hugemmap04 FAIL 2
#* hugemmap05 FAIL 2
#* hugemmap05_1 FAIL 2
#* hugemmap05_2 FAIL 2
#* hugemmap05_3 FAIL 2
#* hugeshmat01 FAIL 11
#* hugeshmat02 FAIL 11
#* hugeshmat03 FAIL 11
#* hugeshmctl01 FAIL 11
#* hugeshmctl02 FAIL 11
#* hugeshmctl03 FAIL 11
#* hugeshmdt01 FAIL 11
#* hugeshmget01 FAIL 11
#* hugeshmget02 FAIL 11
#* hugeshmget03 FAIL 11
#* hugeshmget05 FAIL 11
#* ar FAIL 1
#* ld FAIL 1
#* ldd FAIL 1
#* nm FAIL 1
#* objdump FAIL 1
#* file FAIL 10
#* tar FAIL 2
#* cron FAIL 10
#* logrotate FAIL 1
#* cpio FAIL 1
#* unzip01 FAIL 1
#* gzip01 FAIL 2
#PART2: test with kernel 3.10.y(fce7f0eaa6f7cf5f7873bf2511e33d4bdee914dd) and BSP410 toolchain
#:(the failed cases in part1 are including)
#* swapoff01 FAIL 2
#* swapoff02 FAIL 2
#* swapon01 FAIL 2
#* swapon02 FAIL 4
#* swapon03 FAIL 1
#** For swapon01~03, swapoff01,02:
#** Cannot do swapon on a file located on a tmpfs filesystem
#** Re-test when SD card driver is on
#* madvise02 FAIL 1
#** WILLNEED is set and no file is mapped in the memory region
#** In Linux 3.10, no error is returned -> LTP2016 has fixed this issue
#* proc01 FAIL 1
#** Get EAGAIN when accessing "/proc/self/net/rpc/use-gss-proxy“. LTP issue.
#* BindMounts FAIL 1
#** Cannot execute “mount --bind folderA folderB” on a rootfs filesystem
#** The MS_NOUSER flag in rootfs file system is set
#** It’s PASS when /tmp is mounted on tmpfs
#* gf18 FAIL 1
#** It will run failed if tmp is mount on tmpfs sometimes
#** Please ref to https://jira.automotivelinux.org/browse/SPEC-251
#** It can run pass if /tmp is not mount on tmpfs
#PART3: test with kernel 3.10.y(fce7f0eaa6f7cf5f7873bf2511e33d4bdee914dd) and uclibc toolchain
#:(the failed cases in part1 and part2 are including)
#* fstatfs02_64, statfs02_64: uclibc does not check whether the input parameter, buf, is valid or not before accessing it

*The following 4 testcases will fail if these testcase are accessed by NFS with root_squash permission.
** execve02
** mknodat01
** move_pages03
** move_pages11
*Solutions:
** Move these testcases on SD card
** Modify NFS access permission:
**# Cat /etc/exports on LTP server
**# Look for the folder which contents LTP testsuite in /etc/exports
**# Replace the option "root_squash" with "no_root_squash" (need root/sudo permission)
**# Update /etc/exports (command on Ubuntu: sudo exportfs -r)

*The following testcases will fail by permission denied to access program Deploy/opt/ltp/testcase/bin/test3
** execve04
*Solutions:
** change permission: chmod 777 test3
*** before change: -rwxrwxr-x
*** after change: -rwxrwxrwx

*In 3.10 and 4.13, the following 2 testcases will fail if kernel is configured to 8k page
** vma03 - The failed reason is the same as before (kernel 3.4.35)
**: It will fail in 8KB page size configuration. Due to "if (pgoff + (new_len >> PAGE_SHIFT) < pgoff)" in mm/mremap.c will not overflow. pgoff will be 0x7fffffff in 8KB page size config and it will be 0xfffffff in 4KB page size config.
** mmap01 - this case is pass if CONFIG_CPU_CACHE_NONALIASING is set

== 20130104 ==

test with kernel 3.4.35

'''syscalls'''

* execve02 - The setting is CONFIG_HZ=100 , so need to execute "./getrusage04 -m 9"
* getcontext01 - getcontext failed: TEST_ERRNO=???(38): Function not implemented. Not supported in glibc.
* getrusage03 - Not enough memory. Allocating 100Mb is more than my system memory limitation.
* getrusage04 - The target will run chomd syscall to modify server file. It's because there is not permission on the LTP NFS directory
* ioctl03 - It need to add the "802.1d Ethernet Bridging" and "Universal TUN/TAP device driver support" support.
* mkdir09 - This testcase may fail sometimes because the execution timeout is setting to short. You may extend the timeout value for signal to 20 seconds(5 seconds is the original value) and it will pass all the time.
** ./testcases/bin/mkdir09 -t 20
** test_time = 5; /* 0 = run forever or till signal */ <- set this value to 20
* move_pages03 - The target will run chomd syscall to modify server file. It's not permission.
* move_pages11 - The target will run chomd syscall to modify server file. It's not permission.
* mq_notify01 - function not implemented
* mq_open01 - function not implemented
* mq_timedreceive01 -function not implemented
* mq_timedsend01 - function not implemented
* mq_unlink01 - function not implemented
* mknodat01 - It's because there is not permission on the LTP NFS directory. It will pass, for example, when executing on the SD card.
* syscall01 - Turned on CONFIG_UID16 to support testcase request
* remap_file_pages01 - Fix the cache_pages to support not enough 4M ramdisk space
* sysctl03 - this test's results are based on potentially undocumented behavior in the kernel. read the NOTE in the source file for more details
* sysconf01 - This failure was caused by -O2 option. Nothing to do with `actual' variable.
* syslog01~10 - Our ramdisk doesn't have /etc/syslog.conf and /etc/init.d/syslog so the tests will fail.
* utimensat01 - The testcase will execute utimensat_tests.sh script, it include that the useradd and sudo command are not supported in busybox.

'''sched'''
* hackbench - Reduce number of processes in hackbench01 to avoid the out of memory

'''mm'''
* mtest06_2 - Cannot allocate mush memory than physical memory.
* mmapstress06 - for this test to run, it needs a mmap space of 32769 pages
* thp02 - Andes doesn't support hugepages policy.
* max_map_count - Linux Kernel get wrong maps information from toolchains.
* vma01 - This test program clones VMA and checks /proc/$pid/maps file, on an unpatched kernel, there is a single 6*ps VMA for the child.
* vma03 - The New Linux will be pass, because the New kernel will modify the mremap() syscall policy.
** It will fail in 8KB page size configuration. Due to "if (pgoff + (new_len >> PAGE_SHIFT) < pgoff)" in mm/mremap.c will not overflow. pgoff will be 0x7fffffff in 8KB page size config and it will be 0xfffffff in 4KB page size config.
** It seems a kernel bug. We may use this fix "if ((pgoff + (new_len >> PAGE_SHIFT)) << (PAGE_SHIFT-12) < (pgoff << (PAGE_SHIFT-12)))" in mm/mremap.c to fix this issue.
* mm01 - mm01 is OK when the available size of /tmp is greater than 40960000byte (40mb is OK)

'''fs'''
* rwtest*- shell script mode=/bin/bash . After using bash to execute shell script, these tests pass. In addition, the /tmp folder must be mounted after mounting LTP folder. If the /tmp is mounted before LTP folder, the testcase is failed due that the "df" command which is used in testcase does not work exactly. The detail can be found in BUG9687.
* writetest01 - the program is linked with -lpthread, and our pthread Default stack size = 2093056, when the process is running, it use stack above 4m, it will cause segmentation fault, so the solution is ignore -lpthread or decreasing BLOCKSIZE
* fs_di - fs_di is OK when the available size of /tmp is greater than 90mb
* splice02 - splice02 is OK when the available size of /tmp is greater than 75.2mb

'''fs_bind'''

The testcase is failed and handed due that the mount and diff in busybox.

A. Hanged reason
*The following two reasons cause that the fs_bind is hanged.
** 1. In busybox 1.22, the -B option of diff does not work exactly.
** 2. In addition, the command 'set -o pipefail' does not work if we use /bin/bash as shell.

*Using the following solutions can solve the hanged reason.
**Solution for 1. Replacing the busybox's diff by GUN diff utilities
**Solution for 2. Workaround by modifying the code in test_fs_bind.sh. The new code is in the following.
grep_proc_mounts()
{
local rc=0
grep -F "${sandbox}" /proc/mounts >/dev/null 2>&1
rc=$?
grep -F "${sandbox}" /proc/mounts | sort -b
return $rc
}
grep_mounts()
{
local rc=0
mount | grep -F "${sandbox}" > /dev/null 2>&1
rc=$?
mount | grep -F "${sandbox}" | sort -b
return $rc
}
B. Failed reason
*The Failed case can be reduced to the following case
A. Mount SD on /tmp
B. Create directory a, b in /tmp
C. mount --bind /tmp/a /tmp/b
D. umount /tmp/a
-> show:"umount: can't umount /tmp/a: Invalid argument"
E. umount /tmp/b
-> OK
*The umount instruction does not work in the busybox (version 1.22)
** (p.s) I build the busybox by x86 toolchain and use the busybox to re-run the above flow. The result is the same

'''admin_tools'''
*su01,cron02,cron_deny01,cron_allow01,cron_dirs_checks01,at_deny01,at_allow01,acl_test01:not support useradd,userdel in current busybox,so it will fail.

'''hugetlb'''
*Hugetlb subset - Currently not supporting HUGETLLB_PAGE and HUGETLBFS.

'''numa'''
*Numa-testcases - Andes arch does not support NUMA, it will fail but it is false positive

'''commands'''
*ar - Several options are not supported in ar.
*ld - Need native ld.
*ldd - Currently we don't support file, native cc and native ldd.
*nm01 - Currently we don't support native cc.
*objdump - Need native objdump.
*file - file command is not supported in current busybox.
*logrotate - logrotate command is not supported in current busybox.
*gzip01 - busybox implementation of gzip and gunzip don't take -r option.
*tar - The sub-testcases, tar03 and tar05, use the "file" command but it is not supported in busybox. Regardless of "file" command, the testcase is pass.
*unzip - After executing unzipping, the result is not same with the expected results. However, the unzipping result is the same with X86 when using the same busybox. Therefore, we think it is false positive.
*cpio- The testcase is pass after adding the "-H newc" option. In addition, The testcase uses the "file" command but it is not supported in busybox
*cron - crond does not be trigged after executing crontab <file>. The test case will pass by modifying the following items.
**a. It needs to create directory 'mkdir -p /var/spool/cron/crontabs'. The task list of crontab is stored in this file.
**b. In current busybox, the crond does not execute automatically after crontab. The command "crond" has to be entered after "crontab <file>" command
**c. Because of unworkable corntab, the keywords of grepping "crontab, REPLACE" should be replaced by "crond"
**d. The keywords for grepping "CMD" should be replaced by "cmd"
**e. "Crontab -r" only remove the /var/spool/cron/crontabs/root file. It cannot stop the crond process. Therefore, the test5# can not work.

== 20110915 ==
test with kernel 2.6.32 f5af60a9f90e0bf

'''syscalls'''

* accept04_01 - The accept4() system call is available starting with Linux 2.6.28; support in glibc is available starting with version 2.10. Currently on x86 and SPARC support it.
* add_key* - syscall __NR_add_key not support
* cacheflush01 - syscall __NR_cacheflush not supported (the same with x86)
* creat08 - On NFS, if you open a file in a empty directory and then unlink the file, you can't rmdir this empty dir.(the same with x86)
* fallocate01~03 - The old commit should be unnecessary to support nds architecture. But support is provided by glibc since version 2.10. Can't pass in ext4 now.
* fcntl24~26 - It can't run on ramdisk. After umount /tmp(ramdisk), the test is passed
* fork13 - It takes a lot of time to complete this test, but it will pass
* ftruncate04 - The filesystem where /tmp is mounted does not support mandatory locks. Cannot run this test. (the same with x86)
* getcontext01 - getcontext failed: TEST_ERRNO=???(38): Function not implemented. Not supported in glibc.
* getcpu01 - x86 only
* get_robust_list01 - syscall __NR_get_robust_list not supported
* getrusage03 - memory is not enough
* ioctl03 - Current kernel defconfig doesn't turn on CONFIG_TUN. If one wants to test it, turn on CONFIG_TUN in kernel and create the /dev/net/tun device node. (c/10/200) (the same with x86)
* ioperm01~02,iopl01~02 - LSB v1.3 does not specify the syscall for this architecture.
* modify_ldt01~02 - Only for i386.
* mq_notify01 - function not implemented
* mq_open01 - function not implemented
* mq_timedreceive01 -function not implemented
* mq_timedsend01 - function not implemented
* mq_unlink01 - function not implemented
* ppoll01 - syscall __NR_ppoll not supported
* profil01 - strange ADDRESS_OFFSET calculation, and the code to disable profil seems not to work
* ptrace05 - can not be reproduced again.
* quotactl01 - syscall __NR_quotactl not supported
* remap_file_pages01 - ramdisk space is not enough to create a 4M file, after modifying cache_pages=300 this will pass.
* setrlimit01 - the 4th testcase. After fork, child cause a segment fault, but it register a handler to handle this signal. When return from handler, it will meet segment fault again and again. So we mark the handler to fix it.
* set_robust_list01 - This test fail, but it shows pass. Need to implement futex_atomic_cmpxchg_inatomic in kernel/futex.c to enable futex
* sgetmask01 - syscall __NR_ssetmask not supported
* signalfd4_* - syscall __NR_signalfd4 not supported
* splice02 - kernel with ABI1 support will fetch function arguments from wrong stack when the number of arguments are more than six.
* swapoff*, swapon* - cannot do swap op on a file located on a tmpfs filesystem. Mount sd card to /tmp, this will pass
* sync_file_range01 - Add nds arch support based on old commit.
* sysconf01 - the sysconf returned errno is EINVAL, but the variable 'actual' assigned by errno is not EINVAL, that's because in unistd.h
extern long int sysconf (int __name) __THROW __attribute__ ((__const__));
the value will be optimized.
unistd.h is not consitent with the one in x86
* syslog* - Our ramdisk doesn't have /etc/syslog.conf and /etc/init.d/syslog so the tests will fail.
* utimensat01 - useradd and sudo are not supported in busybox.
* vmsplice01 - Cannot do splice() on a file located on an NFS filesystem. It will pass when using local directory.
* write02 - if /tmp is 20m, the size is not enough.

'''mm'''
* mtest06_2 - mmap2 - main(): mmap(): Cannot allocate memory. We don't have this much memory.
* mm01 - mmap001 -m 10000, mmap a /tmp/xxx file with 10000 page to an array, and write to the array. If /tmp is 20m, it will fail
* mmapstress06 - for this test to run, it needs a mmap space of 32769 pages
* mmapstress08 - x86 only
* oom01, oom03 - test is not designed for 32-bit system.
* thp01 - because we link 'true' to busybox, so the busybox will use arg[0] to see what program will be executed. We need to add array[0] = "true"; to explicitly tell it.
* vma01 - unexpected VMA found. (x86 fail too)

'''ipc'''
* semaphore_test_01 - SEMS=`LANG= ipcs -s | awk '{print $2}' | grep [[:digit:]]` in run_semaphore_test_01.sh should use
SEMS=`LANG= ipcs -s | awk '{print $2}' | grep '[[:digit:]]'` to avoid shell expansion

'''sched'''
* hackbench01 - cherry-pick old commit will solve this

'''fs'''
* rwtest* , iogen - After using bash to execute shell script, these tests pass. In addition, the /tmp folder must be mounted after mounting LTP folder. If the /tmp is mounted before LTP folder, the testcase is failed due that the "df" command which is used in testcase does not work exactly. The detail can be found in BUG9687.
* gf01 - use 50mb tmpfs as /tmp, it will pass , or you can mount sd card to /tmp.
* inode02 - use 50mb tmpfs or use sd card as /tmp, it will pass.
* writetest01 - the program is linked with -lpthread, and our pthread Default stack size = 2093056, when the process is running, it use stack above 4m, it will cause segmentation fault, so 1. can ignore -lpthread or 2. can decrease BLOCKSIZE
* fs_di - create /fs_di225/testfile with size 30mb, our ramdisk is too small,it should allocate upper 30mb size.

== 20090331 ==
* accept04_01 - The accept4() system call is available starting with Linux 2.6.28; support in glibc is available starting with version 2.10. Currently on x86 and SPARC support it.
* ar - Several options are not supported in ar. Option does not support in busybox ls. No diff in the current busybox.
* capset02 - Marked as false positive by LTP.
* cpio01 - busybox implementation of cpio doesn't take -o option.
* cp_tests01 - diff command is not supported in busybox v1.1.2 which is used on and before BSP v2.0.0. diff command in busybox v1.13.2 implementation doesn't support -B option.
* creat08 - It's because there is not permission on the LTP NFS directory. It will pass, for example, when executing on the SD card.
* cron - Currently we don't support crontab.
* dup3_01 - New syscall (dup3), added support in kernel on 20090526, but needs toolchain sync up unistd.h.
* execve04 - failed: expected EMFILE(24). Kernel commit e7b9b550f53e81ea38e71d322d6f95730df058a2 on 20090329 has removed the file get code. The execve syscall will never return EMFILE. execve04 has been remove from LTP on 20090527.
* hugetlb subset - Currently not supporting HUGETLLB_PAGE and HUGETLBFS, now only X86, IA64, PPC64, SPARC64, SH, S39064 does.
* ioctl03 - Current kernel defconfig doesn't turn on CONFIG_TUN. If one wants to test it, turn on CONFIG_TUN in kernel and create the /dev/net/tun device node. (c/10/200)
* sys_epoll02 - Not support in glibc?
* epoll_create1_01 - New syscall (epoll_create1), added support in kernel on 20090526, but needs toolchain sync up unistd.h.
* eventfd2* - New syscall (eventfd2), added support in kernel on 20090526, but needs toolchain sync up unistd.h.
* fallocate - Not supported in ext2.
* fchownat01 - It's because there is not permission on the LTP NFS directory. It will pass, for example, when executing on the SD card.
* file* - file command is not supported in busybox.
* fs_perms - It's because there is not permission on the LTP NFS directory. It will pass, for example, when executing on the SD card.
* getcontext01 - Not supported in glibc.
* get_robust_list01- Kernel not supported. Architectures need to implement the new futex_atomic_cmpxchg_inatomic() inline function before writing up the syscalls (that function returns -ENOSYS right now).
* gzip01 - busybox implementation of gzip and gunzip don't take -r option.
* inotify - Need sys/inotify.h. Not supported in current glibc 2.3.5.
* inotify_init1_01 - New syscall (inotify_init1), added support in kernel on 20090526, but needs toolchain sync up unistd.h.
* lchown02, lchown02_16 broken - Environment setting. http://e-andes.andestech.com/cgi-bin/bugzilla3/show_bug.cgi?id=5603
* ld - Need native ld. Also, diff command is not supported in busybox v1.1.2 which is used on and before BSP v2.0.0. diff command in busybox v1.13.2 implementation doesn't support -B option.
* ldd01 - Currently we don't support file, native cc and native ldd.
* ln_tests01 - diff command is not supported in busybox v1.1.2 which is used on and before BSP v2.0.0. diff command in busybox v1.13.2 implementation doesn't support -B option.
* logrotate - Currently we don't support logrotate command.
* mail - Currently we don't support mail and useradd commands.
* mem01 - May allocate fail. Due to the free memory in /proc/cpuinfo may vary until it is really allcated.
* mem02 -
*# Allocate more than system memory.
*# pm1 != pm2, free did not dealloc memory, free and than allocate is not guaranteeing you get the same address.
* mkdir_tests01 - diff command is not supported in busybox v1.1.2 which is used on and before BSP v2.0.0. diff command in busybox v1.13.2 implementation doesn't support -B option.
* mknodat01 - It's because there is not permission on the LTP NFS directory. It will pass, for example, when executing on the SD card.
* mmap2 - main(): mmap(): Cannot allocate memory. We don't have this much memory.
* mmapstress03 - Shouldn't consider error when returning EINVAL according to mmap manual. Will send patch to LTP.
* mmap01 - Fail if mmap (shared | fixed) not the same color memory.
* mmstress - Child may cause thread manager to exit first and waits for the signal sent by thread manager and thus hang.
* mallocstress - Child may cause thread manager to exit first and waits for the signal sent by thread manager and thus hang.
* mv_tests01 - diff command is not supported in busybox v1.1.2 which is used on and before BSP v2.0.0. diff command in busybox v1.13.2 implementation doesn't support -B option.
* nm01 - Currently we don't support native cc.
* objdump - Need native objdump. Also, diff command is not supported in busybox v1.1.2 which is used on and before BSP v2.0.0. diff command in busybox v1.13.2 implementation doesn't support -B option.
* open10 - It's because there is not permission on the LTP NFS directory. It will pass, for example, when executing on the SD card.
* openat01 - It's because there is not permission on the LTP NFS directory. It will pass, for example, when executing on the SD card.
* pause03 - Child still exists, pause() still active. Synchronization issue of LTP testcase between parent and child. Skip if fail.
* pipe2* - New syscall (pipe2), added support in kernel on 20090526, but needs toolchain sync up unistd.h.
* posix_fadvise - Not supported in glibc until BSP v2.1.0 released toolchain.
* profil01 - Due to the implementation of the testcase, it may fail if static linked.
* remap_file_pages - Not supported in glibc.
* set_robust_list01- Kernel not supported. Architectures need to implement the new futex_atomic_cmpxchg_inatomic() inline function before writing up the syscalls (that function returns -ENOSYS right now).
* splice01 - Cannot do splice on a file located on an NFS filesystem.
* sysconf01 - sysconf correctly failed, but expected errno (22) != actual (-1). Optimization issue of errno, will fail above O2.
* syslog* - Our ramdisk doesn't have /etc/syslog.conf and /etc/init.d/syslog so the tests will fail.
* tar03, tar05 - file command is not supported in busybox.
* unzip01 - unzip command is not turned on in busybox v1.1.2. And also popd command is not supported.
* utimensat01 - useradd and sudo are not supported in busybox.
* test02 ~ test22 - Linux-2.6.32 x86 + busybox is fail
* rwtest* - They should work under /tmp register. In addition, the /tmp folder must be mounted after mounting LTP folder. If the /tmp is mounted before LTP folder, the testcase is failed due that the "df" command which is used in testcase does not work exactly. The detail can be found in BUG9687.
* mmapstress06 - for this test to run, it needs a mmap space of 32769 pages
* mtest06_2 - Cannot allocate mush memory than physical memory.
=== Kernel Dependent ===
* get*_16 - Turned on CONFIG_UID16 to support these deprecated syscalls in kernel on 20090414.
* set*_16 - Turned on CONFIG_UID16 to support these deprecated syscalls in kernel on 20090414.
* stime01 - Kernel bug, Fixed in kernel on 20090721.
* syscall01 - Turned on CONFIG_UID16 to support these deprecated syscalls in kernel on 20090415.
* timerfd - Turned on CONFIG_TIMERFD to support timerfd syscalls after BSP v2.0.0.
* writev - Fail due to we reject if any of the address in the vector is invalid. Fixed in kernel on 20090305.
* lseek03 - Kernel 3.4.35 support the maximum of arg is 4 in lseek(xx,xx,arg). Therefore, the expected errno is not EINVL when arg=3. Fixed in LTP 20130104
* sockioctl01 - LTP bug for kernel 3.4.35. Fixed in LTP 20130104
* inotify01,02 - LTP bug for kernel 3.4.35. Fixed in LTP 20130104

=== cherry-picked in 20090331 ===
So these cases shouldn't happen in this branch.
* clock_gettime03 - Bug of LTP, fixed by LTP on 20090526.
* diotest4 - Bug of LTP, fixed by LTP on 20090623.
* fallocate - Patch to LTP for our ABI. (Need to send patch to LTP to support NDS32 architecture)
* gf* - Bug of LTP, fixed by LTP on 20090403.
* hackbench01 - Memory not enough. Reduced number of process to fix the OOM.
* ioctl01_02 - Bug of LTP, fixed by LTP on 20090406.
* pselect01 - Bug of LTP, should consider the variation. Send patch to LTP on 20090911.
* sync_file_range01- Patch to LTP for our ABI. (Need to send patch to LTP to support NDS32 architecture)
* timer_create04 - Bug of LTP, fixed by LTP on 20090526.
* chmod05, chmod06, chown03, chown04, fchmod05, fchmod06, fchown03, fchown04 -chown failed. Bug of LTP, fixed by LTP on 20090425.
=== fs_bind ===
testcases which cause fail(false positive) and hang:
* move
**test1 test2 test3 test4 test5 test9 test10 test11 test12 test17 test18 test19 test21
* bind
**test1 test2 test3 test4 test5 test9 test10 test11 test12 test13 test14 test15 test16 test17 test18 test19 test20 test21 test22
* rbind
**test1 test2 test3 test4 test5 test6 test7 test8 test9 test10 test11 test12 test13 test14 test15 test16 test17 test18 test19 test20 test21 test22 test25 test26 test27 test28 test29 test30 test31 test32 test33 test34 test35 test36 test37 test38 test39
=== NUMA ===
Andes arch does not support NUMA, it will fail but it is false positive.
=== admin_tools ===
*su01,cron02,cron_deny01,cron_allow01,cron_dirs_checks01,at_deny01,at_allow01,acl_test01:not support useradd,userdel in current busybox,so it will fail.

== 20080831(obsoleted) ==
* capset02 - Marked as false positive by LTP.
* fallocate - Not supported in ext2.
* getcontext01 - Not supported in glibc.
* get_robust_list01- Kernel not supported. Architectures need to implement the new futex_atomic_cmpxchg_inatomic() inline function before writing up the syscalls (that function returns -ENOSYS right now).
* inotify - Need sys/inotify.h. Not supported in current glibc 2.3.5.
* pause03 - Child still exists, pause() still active. Synchronization issue of LTP testcase between parent and child. Skip if fail.
* posix_fadvise - Not supported in glibc.
* profil01 - Due to the implementation of the testcase, it may fail if static linked.
* mem01 - May allocate fail. Due to the free memory in /proc/cpuinfo may vary until it is really allcated.
* mem02 -
*# Allocate more than system memory.
*# pm1 != pm2, free did not dealloc memory, free and than allocate is not guaranteeing you get the same address.
* mmap01 - Fail if mmap (shared | fixed) not the same color memory.
* mmstress - Child may cause thread manager to exit first and waits for the signal sent by thread manager and thus hang.
* mallocstress - Child may cause thread manager to exit first and waits for the signal sent by thread manager and thus hang.
* remap_file_pages - Not supported in glibc.
* set_robust_list01- Kernel not supported. Architectures need to implement the new futex_atomic_cmpxchg_inatomic() inline function before writing up the syscalls (that function returns -ENOSYS right now).
* sysconf01 - sysconf correctly failed, but expected errno (22) != actual (-1). Optimization issue of errno, will fail above O2.

=== Kernel Dependent ===
* setgid02_16 - Turned on CONFIG_UID16 to support these deprecated syscalls in kernel on 20090414.
* setgroups03_16 - Turned on CONFIG_UID16 to support these deprecated syscalls in kernel on 20090414.
* setgroups04_16 - Turned on CONFIG_UID16 to support these deprecated syscalls in kernel on 20090414.
* syscall01 - Turned on CONFIG_UID16 to support these deprecated syscalls in kernel on 20090415.
* writev - Fail due to we reject if any of the address in the vector is invalid. Fixed in kernel on 20090305.

=== cherry-picked in 20080831 ===
So these cases shouldn't happen in this branch.
* creat07 - fails if there is no permission of the LTP directory, e.g., root_squash in NFS setting. Fixed by LTP on 20081113
* execve05 - fails if there is no permission of the LTP directory, e.g., root_squash in NFS setting. Fixed by LTP on 20081211
* fallocate - Patch to LTP for our ABI. (Need to send patch to LTP to support NDS32 architecture)
* getcpu01 - Bug of LTP, we shouldn't test this case since we don't support. Fixed by LTP on 20081015.
* hackbench01 - Memory not enough. Reduced number of process to fix the OOM.
* semaphore_test_01- gawk is needed but not supported in the ramdisk. Patched to LTP on 20080905.
* setfsuid04 - Bug of LTP. Patched to LTP on 20080908.
* setresuid04 - Bug of LTP. Patched to LTP on 20080908.
* setreuid07 - Bug of LTP. Patched to LTP on 20080908.
* shmem_test_06 - LTP's bug, try to attach to space that is used by stack. fixed by LTP on 20081029.
* sync_file_range01- Patch to LTP for our ABI. (Need to send patch to LTP to support NDS32 architecture)

==Backup==
* capset02 - Marked as false positive by LTP.
* creat07 - fails if there is no permission of the LTP directory, e.g., root_squash in NFS setting. Fixed by LTP on 20081113
* execve02 - May fail due to concurrency issue, patched by LTP on 20080422.
* execve05 - fails if there is no permission of the LTP directory, e.g., root_squash in NFS setting. Fixed by LTP on 20081211
* fallocate - Need to send patch to LTP for our ABI, before that, please bypass these testcases.
* getcontext01 - Not supported in glibc.
* getcpu01 - Bug of LTP, we shouldn't test this case since we don't support. Fixed by LTP on 20081015.
* get_robust_list01- Kernel not supported. Architectures need to implement the new futex_atomic_cmpxchg_inatomic() inline function before writing up the syscalls (that function returns -ENOSYS right now).
* hackbench01 - Memory not enough.
* inotify - Need inotify.h which is in 2.6 header file. Not supported in current toolchain.
* pause03 - Child still exists, pause() still active. Synchronization issue of LTP testcase between parent and child. Skip if fail.
* posix_fadvise - Not supported in glibc.
* profil01 - Due to the implementation of the testcase, it may fail if static linked.
* mem01 - May allocate fail. Due to the free memory in /proc/cpuinfo may vary until it is really allcated.
* mem02 -
*# Allocate more than system memory.
*# pm1 != pm2, free did not dealloc memory, free and than allocate is not guaranteeing you get the same address.
* mmap01 - Fail if mmap (shared | fixed) not the same color memory.
* mmstress -
*# Bug of LTP testcase, fixed by LTP on 20070517.
*# Child may cause thread manager to exit first and waits for the signal sent by thread manager and thus hang.
* mallocstress - Child may cause thread manager to exit first and waits for the signal sent by thread manager and thus hang.
* pselect01 - Sleep time was incorrect:5 != 6, fixed by LTP on 20080703.
* remap_file_pages - Not supported in glibc.
* semaphore_test_01- gawk is needed but not supported in the ramdisk. Patched to LTP on 20080905.
* setfsuid04 - Bug of LTP. Patched to LTP on 20080908.
* setgid02_16 - We don't support these deprecated syscalls.
* setgroups03_16 - We don't support these deprecated syscalls.
* setgroups04_16 - We don't support these deprecated syscalls.
* setresuid04 - Bug of LTP. Patched to LTP on 20080908.
* setreuid07 - Bug of LTP. Patched to LTP on 20080908.
* set_robust_list01- Kernel not supported. Architectures need to implement the new futex_atomic_cmpxchg_inatomic() inline function before writing up the syscalls (that function returns -ENOSYS right now).
* shmem_test_06 - LTP's bug, try to attach to space that is used by stack. fixed by LTP on 20081029.
* shmt09 - The testcase chooses an address that is reserved by architexture, patched to LTP on 20080515.
* sync_file_range01- Need to send patch to LTP for our ABI, before that, please bypass this testcases.
* syscall01 - Linux 2.6 we choice not to support UID16. But we use 2.4 unistd.h that contains UID16 syscall.
* sysconf01 -
*#The testcase doesn't check return value, fixed by LTP on 20080406.
*#sysconf correctly failed, but expected errno (22) != actual (-1). Optimization issue of errno, will fail above O2.
* writev - Fail due to we reject if any of the address in the vector is invalid.

* Cannot open /proc/sys/kernel/pid_max - Not supported in Linux 2.4.
* You need to have lo loopback device up to pass network testcases, connect, send, recv, recvfrom, etc.
* qmm01(mmap001) - It will fail in BSP321. And BSP321 sd driver default configuration is SD1.0. Single test via SD2.0 will pass. How to configure to SD2.0 from SD 1.0. Please refer to http://e-andes/wiki/index.php/Linux_device_driver_path#Linux_Drivers_Tests. Detail verification can refer to T:\users\klmnopqr\rick\bugzilla\bug_10871\qmm01-verify-record.ppt

= Fail cases =
=== not found caues ===
* memcg series
**getconf: not found
* mkfs01 series
** mkfs: not found
* cpuset_regression_test FAIL 6
** cpuset_regression_test.sh: line 146: getconf: not found
*su01 FAIL 127
**sh: su01: not found
*cron02 FAIL 2
**userdel: not found
**useradd: not found
*ar FAIL 2
**/bin/sh: ar: not found
*ldd01 FAIL 1
**ldd01: line 64: ldd: not found
*file01 FAIL 2
**file01.sh: line 106: readelf: not found
* cpuhotplug05: sar

=== 32/64-bit issues ===
* sendfile04: 32-bit fails in glibc wrapper sendfile(), false positive.

=== misc ===
* cpuhotplug06: top command format

= OS kernel konwn issue =
* For BSP v1.3 release
<pre>
ID Opened Sev Pri Assignee Reporter Status Resolution Comp TargetM QAContact Summary
------------------------------------------------------------------------------------------------------------------------------------------------------
2486 2008-02-18 cri P1 Roy Lee Roy Lee RESO REMI OS v200711 Chi Yu LTP diotest05 1 FAIL : readv/writev comparision failed
2487 2008-02-18 cri P1 Roy Lee Roy Lee RESO REMI OS v200711 Chi Yu LTP: alarm02 2 FAIL : alarm(4294967295)
2502 2008-02-20 cri P1 Roy Lee Roy Lee RESO INVA OS v200711 Brandy Ho <<2.4 leopard>> LTP hanged after testcase times02
2546 2008-02-26 cri P1 Shawn Lin Brandy Ho RESO INVA OS v200711 [2008-02-22-maintrunk OS] LTP math fail
2574 2008-02-29 cri P1 Eric Tsai Brandy Ho CLOS FIXE OS v200711 [2008-02-27-branch OS] LTP clone04 Fail
2677 2008-03-17 maj P1 Shawn Lin Brandy Ho RESO INVA OS v200711 [2008-03-13-maintrunk]LTP execve02 Fail
</pre>
= LTP auto test script =
Documentation: T:\users\stuv\vincentc\LTP_auto_test_doc.ppt