2nd packet and subsequent packets are sent based on SIGALRM and
it depends on poll(2) returns with EINTR by the signal. However,
currently poll is rump-ified while signals aren't so the signal
doesn't wake up poll and ping6 doesn't work expectedly.
Rump-ifying signals is unsure (nobody does it for now) and the
combination use of signals and poll makes the logic a bit complex.
So let's fix the defect by stopping using signals for packet
transmissions. The new logic is derived from ping(8).
Bonus: ping6 -i 0.01 works as we expect now while the original
didn't work enough fast.
It exited with status 0 when at least one packet was received
and timed out. However, when we specify -c N (while N > 0),
we expect that it fails if it times out before all sent requests
are replied.
For now at least it will refuse to touch it though as it needs to be
taught more abstraction about directory entries; currently it blindly
uses struct direct from ffs and will croak on the lfs64 64-bit inode
numbers.
To create an lfs64 volume, use -w 64. You can also force a 32-bit
volume with -w 32, but this will fail on devices thta are too large.
lfs64 is the default for devices > 1TB. (1TB rather than 2TB because
daddr_t is signed and negative block numbers can cause interesting
complications.)
For now print a warning that the lfs64 format isn't finalized, because
it isn't. For now any lfs64 use should be limited to test data,
benchmarking, and so forth.
add and remove loops. now when you add a new swap file and run
"swapctl -A" it won't give up after the existing one gives EBUSY
and move on to the rest.
so widen it to 'long long'. pass2 uses it for the number of entries in
a directory (IIUC) which does not need to be wider than int, but for
now let's not try to split into two fields. FUTURE...
This is always uint32_t, but having a name for it both makes things
clearer and avoids confusion about whether it should be 32 or 64 bit.
Note: deployed in only one place (that was erroneously tagged
ondisk32) so far.
- compute the maximum file size using LFS_BLKPTRSIZE()
- use the new IINFO in pass 6 instead of uint32_t pointers
- use accessors to read and write indirect blocks
The LFS64 directory entry has a 64-bit inode number. This is stored as
two 32-bit values to avoid inducing 64-bit alignment requirements.
The exposed type for manipulating directory entries is now
LFS_DIRHEADER, following the same convention as e.g. IFILE and SEGUSE.
(But with LFS_ on it, because.)
Also, it turns out that dirhash needs a compile-time-constant version
of LFS_DIRECTSIZ(LFS_MAXNAMLEN+1), independent of 64-vs-32, so create
LFS_MAXDIRENTRYSIZE for this. Sigh.
it in place of (variously) memcpy and strlcpy. (The latter isn't even
correct; was probably changed blindly from strncpy at some point.)
The new function zeroes the padding in the directory entry instead of
leaving trash behind.
0 instead of size LFS_MAXNAMLEN+1, and preparatory to having accessor
functions for d_name. In particular, don't create prototype entries
and copy them, and access the name field only for directory structures
that are in buffers with space for the name to exist.
(lfs_dirheader)
Take the opportunity to improve the directory generation code in
make_lfs.c. (Everything else was unaffected by virtue of using
accessor functions.)
getaddrinfo and if_nametoindex are newly added to interpret interface names
in IPv6 link-local addresses. if_nametoindex looks not used in rump.route,
but it is needed because it is used indirectly from getaddrinfo.
From s-yamaguchi@IIJ
Therefore, storing the value in the superblock and reading it out
again is silly and offers the opportunity for it to become corrupted.
So, don't do that (most of the code already didn't) and use the
existing constant instead. Initialize new 32-bit superblocks with
the value for the sake of old userland programs, but don't keep the
value in the 64-bit superblock at all.
(approved by Margo Seltzer)
Reasoning as before.
Note that I am not going through and checking for 64->32 truncations
in inode numbers; I'm sure there are quite a few, but that's a project
for later.
This prevents regressions in the ulfs code when switching to the new
accessors. Note that while adding byteswapping to the other accessors
is straightforward, I haven't done it yet; and that also is not enough
to make LFS_EI work, because there are places lying around that bypass
the accessors for one reason and another and all of them need to be
updated. That is going to have to wait for a later day as LFS_EI is
not on the critical path right now.
(Mostly.)
The ufs-derived ones are fake structure member macros, which are gross
and not very safe. Also, it seems that a lot of places in the lfs code
were using the ffsv1 branch of them unconditionally, and this way it's
guaranteed all those places have been updated.
Found while doing this: for non-devices, have getattr produce NODEV
in the rdev field instead of leaking the address of the first direct
block.
(This part changes the native lfs code; the ufs-derived code already
has 64 vs. 32 logic, but as aspects of it are unsafe, and don't
entirely interoperate cleanly with the lfs 64/32 stuff, pass 2 will be
rehashing that.)
Add pieces of support for using both superblock types where
convenient, and specifically to the superblock accessors, but don't
actually enable it anywhere.
First substantive step on PR 50000.
in blocks instead of frags, so use lfs_blkstofrags to correct it.
This code was instead multiplying by the block size divided by
DEV_BSIZE to get the number of disk blocks rather than the number of
frags. (I gather that originally these were the same, but they're not
necessarily any more.)
If there are many enabled flags, ifconfig divides flags output into
multiple formatted strings due to snprintb_m, e.g.,
wm0: flags=8b43<UP,BROADCAST,RUNNING,PROMISC,ALLMULTI,SIMPLEX>wm0: flags=8b43<MULTICAST> mtu 1500
This behavior is probably unexpected. The change always outputs enabled
flags at once like this:
wm0: flags=8b43<UP,BROADCAST,RUNNING,PROMISC,ALLMULTI,SIMPLEX,MULTICAST> mtu 1500
This contains all the accessor functions and macros out of lfs.h.
Add an include of lfs_accessors.h after all uses of lfs.h... except
for code that wants to define its own struct lfs-alike that the
accessors are supposed to play along with. For these, set STRUCT_LFS
and include lfs_accessors.h after the necessary structure has been
defined, so that lfs_accessors.h can emit functions in terms of it.
(This changes the rest of the code over; all the accessors were
already added.)
The difference between this commit and the previous one is arbitrary,
but the previous one passed the regression tests on its own so I'm
keeping it separate to help with any bisections that might be needed
in the future.
superblock. This will allow switching between 32/64 bit forms on the
fly; it will also allow handling LFS_EI reasonably tidily. (That
currently doesn't work on the superblock.)
It also gets rid of cpp abuse in the form of fake structure member
macros.
Also, instead of doing sleep/wakeup on &lfs_avail and &lfs_nextseg
inside the on-disk superblock, add extra elements to the in-memory
struct lfs for this. (XXX: these should be changed to condvars, but
not right now)
XXX: this migrates a structure needed by the lfs code in libsa (struct
salfs) into lfs.h, where it doesn't belong, but for the time being
this is necessary in order to allow the accessors (and the various
lfs macros and other goop that relies on them) to compile.
After that chanage, "MAXPARTITIONS" constant is not for the target port.
If host's MAXPARTITIONS is larger than a value of the target label and
target endianness is different from the build host, bswaplabel() could
overwrite data beyond the disklabel and primary boot stored after
LABELSECTOR in images might be corrupted.
This fixes boot failure of sun2 liveimage built by
"build.sh -U -m sun2 release live-image"
on TME.
Should be pulled up to netbsd-7.
Remove sync(2) call before unmount(2) in umount(8). This sync(2) is useless
since unmount(2) will perform a VFS_SYNC anyway.
But moreover, this sync(2) may be harmful, as there are some situation where
it cannot return (unreachable NFS server, for instance), causing umount -f
to be uneffective.
Coincidentally, the change also works around a gcc 5.1 bug which causes
a segmentation fault when trying to compile the longer version (guess
the compiler got exhausted, or something).
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66345
resizedisk will adjust the media size in the headers, whereas
recover simply copies the existing header over the missing one.
XXX recover and resizedisk should probably be merged (even if just
partially, so that recover can properly handle media size changes).
Also, reading in the GPT should probably be centralised so that
error handling can be centralised, and users aren't given misleading
messages (i.e. they aren't told to run recover when they should be
running resizedisk). However, something that can be quickly pulled
up to netbsd-7 was needed, and there isn't time for a major overhaul
or rewrite.
christos