v2.0.x: memory/patcher: do not hook madvise

[hjelmn]

It is not possible to hook madvise at this time due to a deadlock when
using glibc.

Signed-off-by: Nathan Hjelm hjelmn@lanl.gov
(cherry picked from commit 3caeda2)
Signed-off-by: Nathan Hjelm hjelmn@lanl.gov

[jsquyres]

@jladd-mlnx @artpol84 Mellanox Jenkins fails with this error:

19:16:57 [jenkins03:19952] Error: pshmem_init.c:74 - _shmem_init() SHMEM failed to initialize - aborting

Is that a real error, or a false positive?

[jsquyres]

It looks like the Mellanox Jenkins may have pulled in a zillion commits for this PR. That doesn't seem right -- there should be a single commit...?

[artpol84]

bot:mellanox:retest

[Di0gen]

bot:mellanox:retest

[Di0gen]

bot:mellanox:retest

[Di0gen]

The Mellanox's Jenkins pulled right commit,
But on Jenkins page, it printed the diff HEAD vs master. This is the reason of this long commits list.
Settings changed, now it isn't printing the chanlog list.

[jsquyres]

@Di0gen @artpol84 Thanks for fixing. Can you please check the Mellanox Jenkins for #3123, too? It looks like it also pulled a zillion commits. Just to make sure: we have PRs actively being filed against 4 different branches these days:

• v1.10
• master
• v2.0.x
• v2.x

Shortly (within a week or three), we'll have a 5th branch (v3.x), but the activity on the v1.10 branch will likely die off.

[jsquyres]

@hppritcha We agreed on this PR yesterday on the call, so I'm going to merge.