fs/
~~~

o ext3 data=journal mode is bust.  (fix is in progress)

o ext3/htree readdir can return "." and ".." in unexpected order, which
  might break buggy userspace apps.  Ted has a fix planned.

o AIO/direct-IO writes can race with truncate and wreck filesystems.

  o Easy fix is to add an rwsem to the inode.

o hch: devfs: there''''s a fundamental lookup vs devfsd race that''''s only
  fixable by introducing a lookup vs devfs deadlock.  I can''''t see how this is
  fixable without getting rid of the current devfsd design.  Mandrake seems
  to have a workaround for this so this is at least not triggered so easily,
  but that''''s not what I''''d consider a fix..

o viro: fs/char_dev.c needs removal of aeb stuff and merge of cdev-cidr.
  In progress.

o forward-port sct''''s O_DIRECT fixes

o viro: there is some generic stuff for namei/namespace/super, but that''''s a
  slow-merge and can go in 2.6 just fine

o andi: also soft needs to be fixed - there are quite a lot of
  uninterruptible waits in sunrpc/nfs

o trond: NFS has a mmap-versus-truncate problem

kernel/sched.c
~~~~~~~~~~~~~~

o "Persistent starvation"

  http://www.hpl.hp.com/research/linux/kernel/o1-starve.php

  ingo: "basically by calling sleep(1) in an infinite loop you can end up
  expiring yourself.  The testcode (test-starve.c) triggers this.  This is
  solved by going to sub-timeslices.  Which i''''ve got done a few weeks ago and
  it has seen some testing by others as well.

o Overeager affinity in presence of repeated yields

  http://www.hpl.hp.com/research/linux/kernel/o1-openmp.php

  ingo: this is valid.  fix is in progress.

o The "thud.c" test app.  This is a exploit for the interactivity
  estimator.  it''''s unlikely to bite in real-world cases.  Needs watching.
  Can be ameliorated by setting nice values.

o generic interactivity problems need watching.  We''''ve closed down a number
  of items recently without introducing new ones, so i''''m confident this is
  heading in the right direction.

kernel/
~~~~~~~

o Alan: 32bit uid support is *still* broken for process accounting.

  Create a 32bit uid, turn accounting on.  Shock horror it doesn''''t work
  because the field is 16bit.  We need an acct structure flag day for 2.6
  IMHO

  (alan has patch)

o nasty task refcounting bug is taking ages to track down.  (bugzilla ref?)

o viro: core sysctl code is racy.  And its interaction wiuth sysfs

o gettimeofday goes backwards.  Merge up David M-T''''s fixes?

o Daniel Jacobowitz <dan@debian.org>: when CLONE_DETACHED threads were
  removed from /proc several approaches were suggested to let procps find out
  about them and none of them were implemented.  There''''s some real potential
  for badness with these mostly-invisible processes.  Something needs to be
  added so that we can display and detect them.

lib/kobject.c
~~~~~~~~~~~~~

o kobject refcounting (comments from Al Viro):

  _anything_ can grab a temporary reference to kobject.  IOW, if kobject is
  embedded into something that could be freed - it _MUST_ have a destructor
  and that destructor _MUST_ be the destructor for containing object.

  Any violation of the above (and we already have a bunch of those) is a
  user-triggerable memory corruption.

  We can tolerate it for a while in 2.5 (e.g.  during work on susbsystem we
  can decide to switch to that way of handling objects and have subsystem
  vulnerable for a while), but all such windows must be closed before 2.6 and
  during 2.6 we can''''t open them at all.

mm/
~~~

o Overcommit accounting gets wrong answers

  o gets confused by reclaimable-but-not-freed truncated ext3 pages.

o GFP_DMA32 (or something like that).  Lots of ideas.  jejb, zaitcev,
  willy, arjan, wli.

o access_process_vm() doesn''''t flush right.  We probably need new flushing
  primitives to do this (davem?)

modules
~~~~~~~

  (Rusty)

o The .modinfo patch needs to go in.  It''''s trivial, but it''''s the major
  missing functionality vs. 2.4.  Keeps bouncing off Linus.

o __module_get(): "I know I have a refcount already and I don''''t care
  if they''''re doing rmmod --wait, gimme.".  Keeps bouncing off Linus.

o Per-cpu support inside modules (have patch, in testing).

o shemminger: The module remove rework that Rusty and Dave are working on
  needs to be fixed before 2.6.  Right now, it is impossible to write a
  protocol or network device that can be safely unloaded when it is a module.

  See:
        http://www.osdl.org/archive/shemminger/modules.html

  (This is "two stage unload")

net/
~~~~

  (davem)

o UDP apps can in theory deadlock, because the ip_append_data path can end
  up sleeping while the socket lock is held.

  It is OK to sleep with the socket held held, normally.  But in this case
  the sleep happens while waiting for socket memory/space to become
  available, if another context needs to take the socket lock to free up the
  space we could hang.

  I sent a rough patch on how to fix this to Alexey, and he is analyzing
  the situation.  I expect a final fix from him next week or so.

o Semantics for IPSEC during operations such as TCP connect suck currently.

  When we first try to connect to a destination, we may need to ask the
  IPSEC key management daemon to resolve the IPSEC routes for us.  For the
  purposes of what the kernel needs to do, you can think of it like ARP.  We
  can''''t send the packet out properly until we resolve the path.

  What happens now for IPSEC is basically this:

  O_NONBLOCK: returns -EAGAIN over and over until route is resolved

  !O_NONBLOCK: Sleeps until route is resolved

  These semantics are total crap.  The solution, which Alexey is working
  on, is to allow incomplete routes to exist.  These "incomplete" routes
  merely put the packet onto a "resolution queue", and once the key manager
  does it''''s thing we finish the output of the packet.  This is precisely how
  ARP works.

  I don''''t know when Alexey will be done with this.

o There are those mysterious TCP hangs of established state sockets.
  Someone has to get a good log in order for us to effectively debug this.

net/*/netfilter/
~~~~~~~~~~~~~~~~

  (Rusty)

o Handle non-linear skbs everywhere.  This is going in via Dave now.

o Rework conntrack hashing.

o Module relationship bogosity fix (trivial, have patch).

sound/
~~~~~~

o rmk: several OSS drivers for SA11xx-based hardware in need of
  ALSA-ification and L3 bus support code for these.

o rmk: linux/sound/drivers/mpu401/mpu401.c and
  linux/sound/drivers/virmidi.c complained about ''''errno'''' at some time in the
  past, need to confirm whether this is still a problem.

o rmk: need to complete ALSA-ification of the WaveArtist driver for both
  NetWinder and other stuff (there''''s some fairly fundamental differences in
  the way the mixer needs to be handled for the NetWinder.)

  (Issues with forward-porting 2.4 bugfixes.)
  (Killing off OSS is 2.7 material)

global
~~~~~~

o Lots of 2.4 fixes including some security are not in 2.5

o HZ=1000 caused lots of lost timer interrupts.  ACPI or SMM.  (andi,
  jstultz, arjan)

o There are about 60 or 70 security related checks that need doing
  (copy_user etc) from Stanford tools.  (badari is looking into this, and
  hollisb)

o A couple of hundred real looking bugzilla bugs

o viro: cdev rework.  Main group is pretty stable and I hope to feed it to
  Linus RSN.  That''''s cdev-cidr and ->i_cdev/->i_cindex stuff

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

※尚未完成的新特性与加速优化:

From:   Andrew Morton <akpm@digeo.com>
To:   linux-kernel@vger.kernel.org
Subject:   Re: must-fix, version 6
Date:   Fri, 30 May 2003 16:39:07 -0700

The should-fix list, version 6.

Not-ready features and speedups
===============================

Legend:

PRI1: We''''re totally lame if this doesn''''t get in
PRI2: Would be nice
PRI3: Not very important

drivers/block/
~~~~~~~~~~~~~~

o Framework for selecting IO schedulers.  This is the main one really.
  Once this is in place we can drop in new schedulers any old time, no risk.

  PRI1

o Anticipatory scheduler.&n

上一页  [1] [2] [3] [4] [5] [6] [7] [8] [9]  下一页