The tuntap driver allows an admin to create persistent devices and
assign ownership of them to individual users. Unfortunately, relaxing
the permissions on the /dev/net/tun device node so that they can
actually use those devices will _also_ allow those users to create
arbitrary new devices of their own. This patch corrects that, and
adjusts the recommended permissions for the device node accordingly.
Signed-off-By: David Woodhouse <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
The new <linux/dmaengine.h> header shouldn't be included from
the !__KERNEL__ portion of tcp.h
Signed-off-by: David Woodhouse <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
Also, make sure dirents are marked REF_UNCHECKED when we 'discover' them
through eraseblock summary.
Signed-off-by: David Woodhouse <[email protected]>
Failing to do so makes the calculated length of the last node incorrect,
when we're not using eraseblock summaries.
Signed-off-by: David Woodhouse <[email protected]>
Especially when summary code is used, we can have in-memory data
structures referencing certain nodes without them actually being readable
on the flash. Discard the nodes gracefully in that case, rather than
triggering a BUG().
Signed-off-by: David Woodhouse <[email protected]>
Let's not attempt the abolition of mtd->type until/unless it's properly
thought through. And certainly, let's not do it by halves.
Signed-off-by: David Woodhouse <[email protected]>
jffs2_zlib_exit() and free_workspaces() shouldn't be marked __exit because
they get called in the error case from the init functions.
Signed-off-by: David Woodhouse <[email protected]>
Since rb_insert_color() is part of the _public_ API, while the others are
purely internal, switch to be consistent with that.
Signed-off-by: David Woodhouse <[email protected]>
The rest of the file uses these types instead of C99 types.
Acked-by: Dmitry Torokhov <[email protected]>
Signed-off-by: David Woodhouse <[email protected]>
It's used from the initfunc in case of failure too. We could actually do
with an '__initexit' for this kind of thing -- when built in to the
kernel, it could do with being dropped with the init text. We _could_
actually just use __init for it, but that would break if/when we start
dropping init text from modules. So let's just leave it as it was for now,
and mutter a little more about random 'janitorial' fixes from people who
aren't paying attention to what they're doing.
Signed-off-by: David Woodhouse <[email protected]>
This allows us to drop another pointer from the struct jffs2_raw_node_ref,
shrinking it to 8 bytes on 32-bit machines (if the TEST_TOTLEN) paranoia
check is turned off, which will be committed soon).
Signed-off-by: David Woodhouse <[email protected]>
PCI is faked on these devices by SMM traps. Don't depend on that --
check for the chipset directly instead.
Signed-off-by: David Woodhouse <[email protected]>
Print wasted_size in scanned eraseblocks, print range correctly for
summary dirent and inode entries.
Signed-off-by: David Woodhouse <[email protected]>
Preallocation of refs is shortly going to be a per-eraseblock thing,
rather than per-filesystem. Add the required argument to the function.
Signed-off-by: David Woodhouse <[email protected]>
One more place where we were changing the accounting info without
actually allocating a ref for the lost space...
Signed-off-by: David Woodhouse <[email protected]>
Random unthinking 'cleanup' caused debug messages like this:
Obsoleting node at 0x0006daf4 of len 0x3a4: <7>Dirtying
If messages are continuation of an existing line, they don't need
to be prefixed with KERN_DEBUG.
THINK. Or you will be replaced by a small shell script.
Signed-off-by: David Woodhouse <[email protected]>
As the first step towards eliminating the ref->next_phys member and saving
memory by using an _array_ of struct jffs2_raw_node_ref per eraseblock,
stop the write functions from allocating their own refs; have them just
_reserve_ the appropriate number instead. Then jffs2_link_node_ref() can
just fill them in.
Use a linked list of pre-allocated refs in the superblock, for now. Once
we switch to an array, it'll just be a case of extending that array.
Signed-off-by: David Woodhouse <[email protected]>
We don't need the upper layers to deal with the physical offset. It's
_always_ c->nextblock->offset + c->sector_size - c->nextblock->free_size
so we might as well just let the actual write functions deal with that.
Signed-off-by: David Woodhouse <[email protected]>
We'll be using a proper list of nodes in the jffs2_xattr_datum and
jffs2_xattr_ref structures, because the existing code to overwrite
them is just broken. Put it in the common part at the front of the
structure which is shared with the jffs2_inode_cache, so that the
jffs2_link_node_ref() function can do the right thing.
Signed-off-by: David Woodhouse <[email protected]>
In a couple of places, we assume that what's at the end of the
->next_in_ino list is a struct jffs2_inode_cache. Let's check
for that, since we expect it to change soon.
Signed-off-by: David Woodhouse <[email protected]>
Let's avoid the potential for forgetting to set ref->next_in_ino, by doing
it within jffs2_link_node_ref() instead.
This highlights the ugliness of what we're currently doing with
xattr_datum and xattr_ref structures -- we should find a nicer way of
dealing with that.
Signed-off-by: David Woodhouse <[email protected]>
When filing REF_OBSOLETE nodes, we'd add their size to the global
'dirty_size' count, but then to the eraseblock's 'used_size' count.
That's not clever.
Signed-off-by: David Woodhouse <[email protected]>
The previous code wouldn't work correctly on architectures which have a
non-empty MODULE_SYMBOL_PREFIX, and this version is neater if slightly
less optimal in the built-in case.
Signed-off-by: David Woodhouse <[email protected]>
Well, almost. We'll actually keep a 'TEST_TOTLEN' macro set for now, and keep
doing some paranoia checks to make sure it's all working correctly. But if
TEST_TOTLEN is unset, the size of struct jffs2_raw_node_ref drops from 16
bytes to 12 on 32-bit machines. That's a saving of about half a megabyte of
memory on the OLPC prototype board, with 125K or so nodes in its 512MiB of
flash.
Signed-off-by: David Woodhouse <[email protected]>
We can't use jffs2_scan_dirty_space() because it doesn't do any locking; it's
only for use at scan time -- hence the 'scan' in the name.
Also, don't allocate refs while we have c->erase_completion_lock held.
Signed-off-by: David Woodhouse <[email protected]>
We don't allocate this locally any more -- it's given to us and owner by
our caller. Also improve the debug messages a little.
Signed-off-by: David Woodhouse <[email protected]>
Next step in ongoing campaign to file a struct jffs2_raw_node_ref for every
piece of dirty space in the system, so that __totlen can be killed off....
Signed-off-by: David Woodhouse <[email protected]>
If __totlen is going away, we need to pass the length in separately.
Also stop callers from needlessly setting ref->next_phys to NULL,
since that's done for them... and since that'll also be going away soon.
Signed-off-by: David Woodhouse <[email protected]>
Make sure we allocate a ref for any dirty space which exists between nodes
which we find in an eraseblock summary.
Signed-off-by: David Woodhouse <[email protected]>
The incoming ref_totlen() calculation is going to rely on the existence
of nodes which cover all dirty space. We can't just tweak the accounting
data any more; we have to call jffs2_scan_dirty_space() to do it.
Signed-off-by: David Woodhouse <[email protected]>
To eliminate the __totlen field from struct jffs2_raw_node_ref, we need
to allocate nodes for dirty space instead of just tweaking the accounting
data. Introduce jffs2_scan_dirty_space() in preparation for that.
Signed-off-by: David Woodhouse <[email protected]>
For RWCOMPAT and ROCOMPAT nodes, we should still allow the mount to
succeed. Just abandon the summary and fall through to the full scan.
Signed-off-by: David Woodhouse <[email protected]>
If we had to allocate extra space for the summary node, we weren't
correctly freeing it when jffs2_sum_scan_sumnode() returned nonzero --
which is both the success and the failure case. Only when it returned
zero, which means fall through to the full scan, were we correctly freeing
the buffer.
Document the meaning of those return codes while we're at it.
Signed-off-by: David Woodhouse <[email protected]>
We should preserve these when we come to garbage collect them, not let
them get erased. Use jffs2_garbage_collect_pristine() for this, and make
sure the summary code copes -- just refrain from writing a summary for any
block which contains a node we don't understand.
Signed-off-by: David Woodhouse <[email protected]>
The same sequence of code was repeated in many places, to add a new
struct jffs2_raw_node_ref to an eraseblock and adjust the space accounting
accordingly. Move it out-of-line.
Signed-off-by: David Woodhouse <[email protected]>
We were calling ref_totlen() 18 times. Even before that becomes a real
function rather than just a dereference, apparently some compilers still
suck anyway. It'll _certainly_ suck after ref_totlen() becomes more
complicated, so calculate it once and don't rely on CSE.
Signed-off-by: David Woodhouse <[email protected]>
This improves the time to mount 512MiB of NAND flash on my OLPC prototype
by about 4%. We used to read the last page of the eraseblock twice -- once
to find the offset of the summary node, and again to actually _read_ the
summary node. Now we read the last page only once, and read more only if
we need to.
We also don't allocate a new buffer just for the summary code -- we use
the buffer which was already allocated for the scan. Better still, if the
'buffer' for the scan is actually just a pointer directly into NOR flash,
we use that too, avoiding the memcpy() which we used to do.
Signed-off-by: David Woodhouse <[email protected]>
Device node major/minor numbers are just stored in the payload of a single
data node. Just extend that to 4 bytes and use new_encode_dev() for it.
We only use the 4-byte format if we _need_ to, if !old_valid_dev(foo).
This preserves backwards compatibility with older code as much as
possible. If we do make devices with major or minor numbers above 255, and
then mount the file system with the old code, it'll just read the first
two bytes and get the numbers wrong. If it comes to garbage-collect it,
it'll then write back those wrong numbers. But that's about the best we
can expect.
Signed-off-by: David Woodhouse <[email protected]>
I still need this hack to work around the fact that softmac doesn't
attempt to associate when we bring the device up...
Signed-off-by: David Woodhouse <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
We have to pack at least the jint16_t structure, because otherwise it'll
be four bytes in size. Thankfully, we can do that and _not_ pack the
actual node structures, and the compiler still doesn't emit stupid code.
Signed-off-by: David Woodhouse <[email protected]>
fs/jffs2/summary.c: In function ‘jffs2_sum_write_data’:
fs/jffs2/summary.c:658: warning: format ‘%zd’ expects type ‘signed size_t’, but argument 4 has type ‘uint32_t’
Signed-off-by: David Woodhouse <[email protected]>
We all inherited the same error from the original NAND board driver which
got copied and changed. Fix them all at once...
Signed-off-by: David Woodhouse <[email protected]>
If we use __attribute__((packed)), GCC will _also_ assume that the
structures aren't sensibly aligned, and it'll emit code to cope with
that instead of straight word load/save. This can be _very_ suboptimal
on architectures like ARM.
Ideally, we want an attribute which just tells GCC not to do any
padding, without the alignment side-effects. In the absense of that,
we'll just drop the 'packed' attribute and hope that everything stays as
it was (which to be fair is fairly much what we expect). And add some
paranoia checks in the initialisation code, which should be optimised
away completely in the normal case.
Signed-off-by: David Woodhouse <[email protected]>
Make it work even with compilers which lack the wit to notice that
THIS_MODULE is always non-NULL. Use #ifdef MODULE instead. It's only
a temporary debugging check anyway.
Signed-off-by: David Woodhouse <[email protected]>
The physmap platform driver conversion added to physmap.c an include
of asm/mach/flash.h which is 1) ARM-specific; and 2) isn't actually
necessary. Remove it.
Signed-off-by: Lennert Buytenhek <[email protected]>
Signed-off-by: David Woodhouse <[email protected]>
We currently get fairly poor behaviour with files which get many short
writes, such as system logs. This is because we end up with many tiny
data nodes, and the rbtree gets massive. None of these nodes are
actually obsolete, so they are counted as 'clean' space. Eraseblocks can
be entirely full of these nodes (which are REF_NORMAL instead of
REF_PRISTINE), and still they count entirely towards 'used_size' and the
eraseblocks can sit on the clean_list for a long time without being
picked for GC.
One way to alleviate this in the long term is to account REF_NORMAL
space separately from REF_PRISTINE space, rather than counting them both
towards used_size. Then these eraseblocks can be picked for GC and the
offending nodes will be garbage collected.
The short-term fix, though -- which probably makes sense even if we do
eventually implement the above -- is to merge these nodes as they're
written. When we write the last byte in a page, write the _whole_ page.
This obsoletes the earlier nodes in the page _immediately_ and we don't
even need to wait for the garbage collection to do it.
Original implementation from Ferenc Havasi <[email protected]>
Signed-off-by: David Woodhouse <[email protected]>
We used to calculate the number of chips to be zero, allocate an array
of that size, then nasty things would happen when we attempt to access
the first object in that zero-sized array.
Now, if the number of _full_ chips that would fit into the map is zero,
we allocate an array of one anyway, and then artificially reduce the
total size of the resulting MTD device to fit in the map.
Signed-off-by: David Woodhouse <[email protected]>
The _board_ driver needs to be mtd->owner, and it in turn pins the
nand.ko module. Fix them all to actually do that, and fix nand.ko not to
overwrite it -- and also to check that the caller sets it, if the caller
is a module.
Signed-off-by: David Woodhouse <[email protected]>
We were scanning for 0xFF through the entire chip -- which takes a while
when it's a 512MiB device as I have on my current toy. The specs only say
we need to check certain bytes -- so do only that.
Signed-off-by: David Woodhouse <[email protected]>
- Implement HW ECC support,
- Provide read_buf() and write_buf() routines using memcpy
- Use on-flash bad block table
- Fix module refcounting
- Avoid read/modify/write in hwcontrol()
- Minor cosmetic fixes
Partly based on code and ideas from Tom Sylla <[email protected]>
Signed-off-by: David Woodhouse <[email protected]>
Oops. Stupid StudlyCaps. Again.
This driver is doubly-deprecated because is was subsumed into doc2000.c
and _also_ we want people to start using the new NAND wrapper for these
devices anyway. But ISTR there was still one person using it because
something didn't work for them. Must chase that up and then I can kill
this.
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: David Woodhouse <[email protected]>
This lacks hardware ECC support and a few optimisations we're going to
want fairly soon, but it works well enough to mount and use JFFS2.
Signed-off-by: David Woodhouse <[email protected]>
Also remove Bob Burroughs' email address, since it's no longer valid.
Acked-by: Eric D Rossman <[email protected]>
Signed-off-by: David Woodhouse <[email protected]>
Fix the conditions under which we poke at the APHY registers in
bcm43xx_phy_initg() to avoid a machine check on chips where they don't
exist.
Signed-off-by: David Woodhouse <[email protected]>
Signed-off-by: John W. Linville <[email protected]>
David Woodhouse