========================================================================== The checksum's (found through sum -r) of the files that you have received (other than this README) are as follows: 52229 8 patchSG0002445 46500 1693 patchSG0002445.eoe1_sw 52073 20 patchSG0002445.idb 60998 847 patchSG0002445.nfs3_sw 08308 875 patchSG0002445.nfs_sw 53866 15 patchSG0002445.trix_eoe_trix_sw ========================================================================== - 1 - 1. Patch_SG0002445_Release_Note This release note describes patch SG0002445 to IRIX 5.3 and IRIX 5.3 with XFS. 1.1 Supported_Hardware_Platforms This patch contains bug fixes for the following machine types. The software cannot be installed on other configurations: o Challenge and Onyx with R4400 processors o Crimson (4D/510) o PowerSeries (4D/120, 4D/2xx, 4D/3xx and 4D/4xx) o Indigo and Indigo2 o Indy 1.2 Supported_Software_Platforms This patch contains bug fixes for IRIX 5.3 and IRIX 5.3 with XFS. It subsumes patches SG0000213, SG0000220, SG0000269, SG0000324, SG0000399. SG0000401, SG0000481, SG0000547, SG0000920, SG0001102, and SG0001477. The software cannot be installed on other configurations. 1.2.1 Warning_for_5.3_with_XFS_configurations Because of some additions to the xfsstubs.a image shipped as a part of this bug, upon relinking the kernel after installing this patch several warnings will be printed out by the linker. They are all along the lines of: Warning: xfsstubs.a(xfsstubs.o): xfs_vfsops: multiply defined previous (used) definition from 'xfs.a'; new (ignored) definition from 'xfsstubs.a' These warnings are harmless and to be expected. They will not affect the correctness of the kernel output by the linker, and they should simply be ignored. 1.3 Bugs_Fixed_by_Patch_SG0002445 This patch contains fixes for the following bugs in IRIX 5.3. Bug numbers from Silicon Graphics bug tracking system are included for reference. - 2 - o The NFS server code is dependent on the structure of the XFS inode. When this changes, as it did in patch SG0000307, the code in the NFS server breaks. (Bug #266670) o The NFS3 server code contains a bug which causes all of its writes to underlying XFS file systems to fail. This error is reported on the console of the client which attempted to flush the data to the server, but it is not reported to the application which performed the write. This often causes the data being written to be lost. (Bug #260821) o Crash in the NFS server kernel routine "fhtovp()" after a CDROM file system has been unmounted. This fixes a bug which allows an NFS server to sometimes cache a kernel pointer to an NFS exported CDROM file system which has been unmounted. The symptom is: you mount and export a CDROM for a remote client, sometime later the media unmounts it, you subsequently crash when the next NFS request arrives for this unmounted file system. (Bug #252076) o NFS server reads to local XFS filesystems are slower than equivalent reads for EFS filesystems. (Bug #253013) This patch changes the NFS server code so that performance on EFS and XFS filesystems is roughly equivalent. o This patch brings exportfs into compliance with its man page regarding the order in which the netgroup and hosts databases are checked for clients in the access list. (Bug #253408) o Makedepend files were getting corrupted when building across NFS (Bug #255451) o 5.3 xfsstubs.a is insufficient (Bug #267202) o Automount daemon can hang (deadlock) permanently if the Clearcase package is installed and in use. (Bug #250595) o DEC OSF/1 clients writing in asynchronous mode to an SGI server could receive a bad error return which caused them to terminate the current write and truncate the file, without informing the client program. This patch corrects the server bug that caused the bad error to be returned. (Bug #266915) - 3 - o SGI NFS3 clients could mishandle an append flag in such a way as to cause a client program to write to the wrong offset in a file opened with O_WRONLY|O_APPEND mode. (Bug #262713) o A header file incompatibility between IRIX4 and IRIX5 caused some GETLK lock requests to be sent to the server as set lock requests. (Bug #257521) o NFS2 mounts could produce incorrect access to files when one of two hard linked files was removed. (Bug #228720, 255462) o Once a server disk filled up, and this was detected when writing to a given file, 5.3 client writes to that file would return errors, even after some space was freed up, as long as that file remained open. (Bug #251853) o File locks are not cleaned up by mail-reader (Bug #251908) o NFS3 now conforms to the 64-bit ABI for UIO_SYSSPACE transfers. (Bug #241490) o NFS panics in rfscall called from nfs_getattr (Bug #248006) o When using an SGI NFS server with an IBM AIX client that uses the "acl" mount option (such as AIX 3.2.5) the "Out of memory" error message is returned unless the "noacl" option is used on the mount. Solaris 2.5 clients will have a similar problem when using the NFS3 protocol (but different symptoms). The fix is to have the SGI server return the "Program unavailable" error when these clients try to use their NFS extensions. (Bug #265008) o Certain corrupted packets could cause an NFS server to crash. The fix is to check the size of the RPC credentials before decoding them. This fix was supposed to be in patch 216, but was accidentally left out. (Bug #251844) o NFS3 client code now correctly interprets quota error return codes sent by the server. (Bug #266893) o Systems that installed patch401 experienced problems with readdir in large directories (greater than two hundred directory entries). This is corrected here. (Bug #272501) - 4 - o File open with Exclusive mode could leave a file with bad permissions, and unreadable. (Bug #280879) o A readdirplus operation could cause a server crash with a double free of allocated kernel memory. This fixes that bug. (Bug #268583) o File read performance over NFS3 was slow until the first time a file was touched. This fixes that problem. (Bug #278593) o Fixes in previous patches impacted NFS3 read performance. This fix restores the original performance for NFS3. There remains one issue with read performance of NFS2 files opened with O_CREATE mode. The work-around is not to use O_CREATE mode if a file is only intended to be read. (Bug #276468) o On certain systems with patch 401 or patch 481 installed, NFS3 write performance was reduced. This fix gets most of the performance back. (Bug #275897) o Certain file handles could provoke a server crash by producing an illegal lookup in export code. This patch fixes that. (Bug #277641) o A bug in setting attributes could cause the 'jot' command to produce a null file. (Bug #307974). o Against DEC servers, a client process could go into an unkillable loop setting file attributes. (Bug #304925). o A bug in setting attributes could cause the 'rcp' command to produce corrupt output. (Bug #300921). o A bug in commit handling could cause mail clients to corrupt a mail file in an nfs3-mounted mail directory. (Bug #311463). o The SGI nohide lookup code was added, to bring nfs3 to the same level as nfs2. (Bug #259503). o Readdir xdr code was enhanced to allow long directory cookies to be decoded correctly. (Bug #312569). o nfs_getattr() locking performance improvements (Bug #318090) o Repeated 'access' calls no longer lead to repeated lookups. (Bug #268702) Caution; this fix addresses a - 5 - particular problem that shows up when a program does repeated 'access' calls. The fix essentially restores the 4.0.5 behaviour, but is probably not relevant for most customers. o NFS V3 client writes sometimes fail to NetApp server. (Bug #328683) o cp across nfs3 over hppi will hang in kernel in an unkillable state, can even hang the system. (Bug #335421) o When using a DMAPI-managed file system via NFS, DMAPI read events were not generated if the read was of a certain size and alignment. (Bug #340913) o When using a Network Appliance server, nulls are sometimes appended to files upon writes. (Bug #333717) o When a client is reading a file that is being written by the server, it occasionally gets invalid (null) data. (Bug #353648) o An NFS version 3 client can cause a panic in the server by looking up a namefs mount point. (Bug #361974) o NFS version 3 writes would cause bio3d processes to consume unreasonable amounts of cpu time and would effectively hang the system until the write finished. (Bug #370725) o The command ls -l would report incorrect sizes for files larger than two gigabytes (Bug #371871) o Server does not notify client of out-of-space error correctly (Bug #397471) o Patch 1348 requires nfs3.a to configure kernel. (Bug #398671) o NFS file locking fails in environments where some SGI systems are running 6.2 or higher. (Bug #489374) 1.4 Subsystems_Included_in_Patch_SG0002445 This patch release includes these subsystems: o patchSG0002445.nfs_sw.nfs o patchSG0002445.eoe1_sw.unix - 6 - o patchSG0002445.nfs3_sw.nfs3 o patchSG0002445.trix_eoe_trix_sw.mac 1.5 Installation_Instructions Because you want to install only the patches for problems you have encountered, patch software is not installed by default. After reading the descriptions of the bugs fixed in this patch (see Section 1.3), determine the patches that meet your specific needs. If, after reading Sections 1.1 and 1.2 of these release notes, you are unsure whether your hardware and software meet the requirements for installing a particular patch, run inst. Patch software is installed like any other Silicon Graphics software product. Follow the instructions in your Software Installation Administrator's Guide to bring up the miniroot form of the software installation tools. Follow these steps to select a patch for installation: 1. At the Inst>prompt, type install patchSGxxxxxxx where xxxxxxx is the patch number. 2. Select the desired patches for installation. 3. Initiate the installation sequence. Type Inst> go 4. You may find that two patches have been marked as incompatible. If this occurs, you must deselect one of the patches. Inst> keep patchSGxxxxxxx where xxxxxxx is the patch number. 5. After completing the installation process, exit the inst program by typing Inst> quit - 7 - To remove a patch, use the versions remove command as you would for any other software subsystem. The removal process reinstates the original version of software unless you have specifically removed the patch history from your system. versions remove patchSGxxxxxxx where xxxxxxx is the patch number. To keep a patch but increase your disk space, use the versions removehist command to remove the patch history. versions removehist patchSGxxxxxxx where xxxxxxx is the patch number.