81 lines · plain
1===================================================2Notes on the change from 16-bit UIDs to 32-bit UIDs3===================================================4 5:Author: Chris Wing <wingc@umich.edu>6:Last updated: January 11, 20007 8- kernel code MUST take into account __kernel_uid_t and __kernel_uid32_t9 when communicating between user and kernel space in an ioctl or data10 structure.11 12- kernel code should use uid_t and gid_t in kernel-private structures and13 code.14 15What's left to be done for 32-bit UIDs on all Linux architectures:16 17- Disk quotas have an interesting limitation that is not related to the18 maximum UID/GID. They are limited by the maximum file size on the19 underlying filesystem, because quota records are written at offsets20 corresponding to the UID in question.21 Further investigation is needed to see if the quota system can cope22 properly with huge UIDs. If it can deal with 64-bit file offsets on all 23 architectures, this should not be a problem.24 25- Decide whether or not to keep backwards compatibility with the system26 accounting file, or if we should break it as the comments suggest27 (currently, the old 16-bit UID and GID are still written to disk, and28 part of the former pad space is used to store separate 32-bit UID and29 GID)30 31- Need to validate that OS emulation calls the 16-bit UID32 compatibility syscalls, if the OS being emulated used 16-bit UIDs, or33 uses the 32-bit UID system calls properly otherwise.34 35 This affects at least:36 37 - iBCS on Intel38 39 - sparc32 emulation on sparc6440 (need to support whatever new 32-bit UID system calls are added to41 sparc32)42 43- Validate that all filesystems behave properly.44 45 At present, 32-bit UIDs _should_ work for:46 47 - ext248 - ufs49 - isofs50 - nfs51 - coda52 - udf53 54 Ioctl() fixups have been made for:55 56 - ncpfs57 - smbfs58 59 Filesystems with simple fixups to prevent 16-bit UID wraparound:60 61 - minix62 - sysv63 - qnx464 65 Other filesystems have not been checked yet.66 67- The ncpfs and smpfs filesystems cannot presently use 32-bit UIDs in68 all ioctl()s. Some new ioctl()s have been added with 32-bit UIDs, but69 more are needed. (as well as new user<->kernel data structures)70 71- The ELF core dump format only supports 16-bit UIDs on arm, i386, m68k,72 sh, and sparc32. Fixing this is probably not that important, but would73 require adding a new ELF section.74 75- The ioctl()s used to control the in-kernel NFS server only support76 16-bit UIDs on arm, i386, m68k, sh, and sparc32.77 78- make sure that the UID mapping feature of AX25 networking works properly79 (it should be safe because it's always used a 32-bit integer to80 communicate between user and kernel)81