* Put the code that runs with the MMU off in a separate section (.start).
* Use a linker script to set the VMA/LMA appropriately for each section.
* Fix kernel page table allocation and mapping of kernel text/data/bss.
...and now that kernel images > 2MB work:
* Add SYMTAB_SPACE to the GENERIC kernel so that we have DDB symbols.
* Use dbsym to load the kernel's DDB symbol area.
set via NFSV3SATTRTIME_TOSERVER and not NFSV3SATTRTIME_TOCLIENT,
add VA_UTIMES_NULL to the va_vflags. This reflects our policy
where we're much more liberal about who can set a & m times to 'now'
than we are about who can set them to a specific time.
Should close PR 15597 from Martin Husemann. Patch is based on the
one Matthias Drochner gave in the PR.
I accidentally committed the previous rev with the wrong commit message
(thanks go to Nick Hudson <skrll@netbsd.org> for picking this up).
Eventually we will probably want to stop descending into utils, but
now is not that time.
policy and the data dependent initialization is very fragile. I.e.
the code assumes because the data entry is of type foo, it has
initialized certain fields and others not.
- make sure list is not NULL before you use it.
- don't realloc on every loop iteration!
- don't increment i when there are short lines; leads to uninitialized
list[i] entries [fixes bills' core-dump, hi bill!]
- don't allocate:
char *p = malloc(len * sizeof(char *))
when you mean:
char *p = malloc(len)
- eliminate duplicated code by creating functions and try not to wrap
- use mvwinnstr instead of mvwinstr
* fdesc_attr(): don't panic for 'unknown' descriptor types, rather use
(*fp->f_ops->fo_stat)() hook, as for DTYPE_SOCKET and DTYPE_PIPE
XXX perhaps use different vnode type than VBAD for these?
* fdesc_setattr(): just return 0 regardless of type, rather than paniccing
for 'unknown' descriptor types
make sure that a failed installation of a rpe-requisite package kills
the attempt to add the package. This reverts pkg_add(1) to traditional
behaviour.
Bump version number accordingly.
