diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2019-11-05 09:44:02 -0800 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2019-11-05 09:44:02 -0800 |
commit | 26bc672134241a080a83b2ab9aa8abede8d30e1c (patch) | |
tree | fe50a086904111be2c3b8ba02489eb0abf2cdc78 /kernel/rcu | |
parent | 7111fa1151e3c66cde3a94250bda72d7144419ff (diff) | |
parent | fa729c4df558936b4a1a7b3e2234011f44ede28b (diff) | |
download | linux-26bc672134241a080a83b2ab9aa8abede8d30e1c.tar.bz2 |
Merge tag 'for-linus-2019-11-05' of git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux
Pull clone3 stack argument update from Christian Brauner:
"This changes clone3() to do basic stack validation and to set up the
stack depending on whether or not it is growing up or down.
With clone3() the expectation is now very simply that the .stack
argument points to the lowest address of the stack and that
.stack_size specifies the initial stack size. This is diferent from
legacy clone() where the "stack" argument had to point to the lowest
or highest address of the stack depending on the architecture.
clone3() was released with 5.3. Currently, it is not documented and
very unclear to userspace how the stack and stack_size argument have
to be passed. After talking to glibc folks we concluded that changing
clone3() to determine stack direction and doing basic validation is
the right course of action.
Note, this is a potentially user visible change. In the very unlikely
case, that it breaks someone's use-case we will revert. (And then e.g.
place the new behavior under an appropriate flag.)
Note that passing an empty stack will continue working just as before.
Breaking someone's use-case is very unlikely. Neither glibc nor musl
currently expose a wrapper for clone3(). There is currently also no
real motivation for anyone to use clone3() directly. First, because
using clone{3}() with stacks requires some assembly (see glibc and
musl). Second, because it does not provide features that legacy
clone() doesn't. New features for clone3() will first happen in v5.5
which is why v5.4 is still a good time to try and make that change now
and backport it to v5.3.
I did a codesearch on https://codesearch.debian.net, github, and
gitlab and could not find any software currently relying directly on
clone3(). I expect this to change once we land CLONE_CLEAR_SIGHAND
which was a request coming from glibc at which point they'll likely
start using it"
* tag 'for-linus-2019-11-05' of git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux:
clone3: validate stack arguments
Diffstat (limited to 'kernel/rcu')
0 files changed, 0 insertions, 0 deletions