While backporting Michael's "pipe: fix limit handling" patchset to a
distro-kernel, Mikulas noticed that current upstream pipe limit handling
contains a few problems:
1 - procfs signed wrap: echo'ing a large number into
/proc/sys/fs/pipe-max-size and then cat'ing it back out shows a
negative value.
2 - round_pipe_size() nr_pages overflow on 32bit: this would
subsequently try roundup_pow_of_two(0), which is undefined.
3 - visible non-rounded pipe-max-size value: there is no mutual
exclusion or protection between the time pipe_max_size is assigned
a raw value from proc_dointvec_minmax() and when it is rounded.
4 - unsigned long -> unsigned int conversion makes for potential odd
return errors from do_proc_douintvec_minmax_conv() and
do_proc_dopipe_max_size_conv().
This version underwent the same testing as v1:
https://marc.info/?l=linux-kernel&m=150643571406022&w=2
v1 (from rfc):
- Re-arrange patchset order, push smaller fixes to the front
- Add a check so that round_pipe_size(size < pipe_min_size) will round
up to round_pipe_size(pipe_min_size) as per man page [RD]
- Add new procfs proc_dopipe_max_size() and helpers to consolidate user
space read / type validation / rounding / assignment [MP]
v2:
- Fix !CONFIG_PROC_SYSCTL build case [buildbots]
v3:
- s/proc_dointvec_minmax/proc_dopipe_max_size/ in comment
preceding pipe_proc_fn()
- Added a fourth patch to address -ERANGE vs. -EINVAL inconsistencies in
do_proc_douintvec_minmax_conv() and do_proc_dopipe_max_size_conv().
Joe Lawrence (4):
pipe: match pipe_max_size data type with procfs
pipe: avoid round_pipe_size() nr_pages overflow on 32-bit
pipe: add proc_dopipe_max_size() to safely assign pipe_max_size
sysctl: check for UINT_MAX before unsigned int min/max
fs/pipe.c | 23 ++++++++++---------
include/linux/pipe_fs_i.h | 1 +
include/linux/sysctl.h | 3 +++
kernel/sysctl.c | 57 ++++++++++++++++++++++++++++++++++++++++++++---
4 files changed, 70 insertions(+), 14 deletions(-)
--
1.8.3.1
