BUGS revision 119679
1 LIST OF KNOWN BUGS IN AM-UTILS OR OPERATING SYSTEMS 2 3 4(1) mips-sgi-irix* 5 6[1A] known to have flaky NFS V.3 and TCP. Amd tends to hang or spin 7infinitely after a few hours or days of use. Users must install recommended 8patches from vendor. Patches help, but not all the time. Otherwise avoid 9using NFS V.3 and TCP on these systems, by setting 10 11 /defaults opts:=vers=2,proto=udp 12 13[1B] yp_all() leaks a file descriptor. Eventually amd runs out of file 14descriptors and hangs. Am-utils circumvents this by using its own version 15of yp_all which uses udp and iterates over NIS maps. The latter isn't as 16reliable as yp_all() which uses TCP, but it is better than hanging. 17 18(I have some reports that older version of hpux-9, with older libc, also 19leak file descriptors.) 20 21[1C] SGI's MIPSpro C compiler on IRIX 6 has the unfortunate habit of 22creating code specificially for the machine it runs on. The ABI and ISA 23used depend very much on the OS version and compiler release used. This 24means that the resulting amd binary won't run on machines different from 25the build host, particularly older ones. Older versions of am-utils 26enforced the O32 ABI when compiling with cc to work around this, but this 27ABI is deprecated in favor of the N32 ABI now, so we use -n32 -mips3 to 28ensure that the binaries run on every host capable of running IRIX 6 at 29all. If this is not appropriate for you, configure with something like 30CC='cc -64' instead to get the desired ABI and ISA. 31 32(2) alpha-unknown-linux-gnu (RedHat Linux 4.2) 33 34hasmntopt(mnt, opt) can go into an infinite loop if opt is any substring 35of mnt->mnt_opts. Redhat 5.0 does not have this libc bug. Here is an 36example program: 37 38#include <stdio.h> 39#include <mntent.h> 40main() 41{ 42 struct mntent mnt; 43 char *cp; 44 mnt.mnt_opts = "intr,rw,port=1023,timeo=8,foo=br,retrans=110,indirect,map=/usr/local/AMD/etc/amd.proj,boo"; 45 cp = hasmntopt(&mnt, "ro"); 46 printf("cp = %s\n", cp); 47 exit(0); 48} 49 50It is possible that sufficiently newer version of libc for RH4.2 fix this 51problem. 52 53 54(3) mips-dec-ultrix4.3 55 56Rainer Orth <ro@TechFak.Uni-Bielefeld.DE> reports 57 58[3A] One needs the Kernel Config Files (UDTBIN430) subset installed to 59compile am-utils, otherwise essential header files (net/if.h, net/route.h, 60rpcsvc/mount.h, rpcsvc/yp_prot.h, rpcsvc/ypclnt.h, sys/proc.h) are 61missing. 62 63[3B] It's probably impossible to build am-utils with DEC C on Ultrix V4.3. 64This compiler is pseudo-ANSI only. Maybe the new ANSI C compiler in V4.3A 65and beyond will do. I successfully used gcc 2.8.1. 66 67[3C] You need to build against a recent libhesiod (I used 3.0.2) and 68libresolv/lib44bsd (I used BIND 4.9.5-P1). The resolver routines in 69libc seem to cause random memory corruption. It is necessary to specify 70LIBS=-l44bsd. lib44bsd is a helper library of libresolv used to supply 71functions like strdup which are missing on the host system. This isn't 72currently autoconfiscated. 73 74[3D] You need to configure with CONFIG_SHELL=/bin/sh5 /bin/sh5 buildall; 75/bin/sh cannot handle the shell functions used in buildall and is both 76buggy and slow. 77 78[3E] At least the gcc 2.7.0 fixincludes-mangled <sys/utsname.h> needs a 79forward declaration of struct utsname to avoid lots of gcc warnings: 80 81RCS file: RCS/utsname.h,v 82retrieving revision 1.1 83diff -u -r1.1 utsname.h 84--- utsname.h 1995/06/19 13:07:01 1.1 85+++ utsname.h 1998/01/27 12:34:26 86@@ -59,6 +59,7 @@ 87 #ifdef KERNEL 88 #include "../h/limits.h" 89 #else /* user mode */ 90+struct utsname; 91 extern int uname _PARAMS((struct utsname *)); 92 #endif 93 #define __SYS_NMLN 32 94 95 96(4) powerpc-ibm-aix4.2.1.0 97 98[4A] "Randall S. Winchester" <rsw@Glue.umd.edu> reports that for amd to 99start, you need to kill and restart rpc.mountd and possibly also make sure 100that nfsd is running. Normally these are not required. 101 102[4B] "Stefan Vogel" <vogel@physik.unizh.ch> reports that if your amq 103executable dump core unexpectedly, then it may be a bug in gcc 2.7.x. 104Upgrade to gcc 2.8.x or use IBM's xlC compiler. 105 106[C] Do not link amd with libnsl. It is buggy and causes amd to core dump 107in strlen inside strdup inside svc_register(). 108 109 110(5) *-linux-rh51 (RedHat Linux 5.1) 111 112There's a UDP file descriptor leak in libnsl in RedHat Linux 5.1. This 113library part of glibc2. Am-utils currently declares redhat 5.1 systems as 114having a "broken yp_all" and using an internal, slower, leak-free version. 115The leak is known to the glibc maintainers and a fix from them is due soon, 116but it is not yet in the glibc-2.0.7-19 RPM. 117 118 119(6) rs6000-ibm-aix4.1.x 120 121A bug in libc results in an amq binary that doesn't work; amq -v dumps core 122in xdr_string. There is no known fix (source code or vendor patch) at this 123time. (Please let amd-dev know if you know of a fix.) 124 125 126(7) *-aix4.3.2.0 127 128The plock() function will pre-reserve all of the memory up to the maximum 129listed in the ulimit. If the ulimit is infinite, plock() will try to take 130all of the system's memory, and fail with ENOMEM (Not Enough Space). 131Normally ulimit may be set to a few gigs of max memory usage, but even that 132is too much; Amd doesn't need more than a few megs of resident memory size 133(depending on the particular usage, number of maps, etc.) Solution: lower 134your ulimit before starting amd. This can be done inside the ctl-amd 135script, but be careful not to limit it too low. Alternatively, don't use 136plock on aix-4.3: set it to plock=no in amd.conf (which is the default if 137you do nothing). 138 139 140(8) *-linux (systems using glibc 2.1, such as RedHat-6.x) 141 142There's a UDP file descriptor leak in the NIS routines in glibc, especially 143those that do yp_bind. Until this is bug fixed, do not set nis_domain in 144amd.conf, but let the system pick up the default domain name as set by your 145system. That would avoid using the buggy yp_bind routines in libc. 146 147 148(9) *-linux (SuSE systems using unfsd) 149 150The user-level nfsd (2.2beta44) on older SuSE Linux systems (and possibly 151others) dies with a SEGV when amd tries to contact it for access to a volume 152that does not exist, or one for which there is no permission to mount. 153 154 155(10) *-*-hpux11 156 157If you're using NFSv3, you must install HP patches PHNE_20344 and 158PHNE_20371. If you don't, and you try to use amd with NFSv3 over TCP, your 159kernel will panic. 160 161 162(11) *-linux* (any system using a 2.2.18+ kernel) 163 164The Linux kernels don't support Amd's direct mounts very well, leading to 165erratic behavior: shares that don't get remounted after the first timeout, 166inability to restart Amd because its mount points cannot be unmounted, 167etc. There are some kernel patches on the am-utils Web site, which solve 168these problems. See http://www.am-utils.org/patches/. 169 170UPDATE: kernels 2.4.10 and later completely disallow the direct mount hack, 171so direct mounts are simply not possible on those Linux kernels. 172 173(12) *-aix5.1.0.0 and *-hpux9* 174 175/bin/sh is broken and fails to run the configure script properly. You need 176to use /bin/ksh instead. The buildall script will do it for you; if for some 177reason you need to run configure directly, run it using 'ksh configure' 178instead of just 'configure'. 179 180[12A] *-aix5.1.* 181 182Apparently there is an NFS client side bug in vmount() which causes amd to 183hang when it starts (and tries to NFS-mount itself). According to IBM 184engineers, this has to do with partial support code for IPv6: the NFS kernel 185code doesn't appear to recognize the sin_family of the amd vmount(), 186although amd does the right thing. The bug appears to have been fixed in 187AIX 5.2. No known fix/patch is available for AIX 5.1 as of now (1/25/2003). 188 189(13) *-linux and *-darwin6.0 190 191Certain linux kernels (2.4.18+ are fine, 2.4.10- are probably bad, those in 192between have not been tested) have a bug which causes them to reconnect 193broken NFS/TCP connections using unprivileged ports (greater than 1024), 194unlike the initial connections which do originate from privileged 195ports. This can upset quite a few NFS servers and causes accesses to the 196mounted shares to fail with "Operation not permitted" (EPERM). 197 198The darwin (MacOS X) kernel defaults to using unprivileged ports, but that 199can be changed by setting the resvport mount flag (which amd sets by 200default). Nonetheless, if a TCP connection breaks, under certain unclear 201circumstances the kernel might "forget" about that flag and start using 202unprivileged ports, causing the same EPERM error above. 203 204 205Erez & Ion. 206 207