HHVM Crashes on file change

[infusion]

Hi,

HHVM segfaults quite often, when I push a new code base with rsync into the docroot. I'm not sure if it's a problem of the stat cache/inotify implementation.

How to reproduce:

• Change some code
• rsync into the docroot
• reload page
  .....

[pid 12935] lseek(26, 1728512, SEEK_SET) = 1728512
[pid 12935] read(26, "setlogin is not implemented and "..., 4096) = 4096
[pid 12935] lseek(26, 1732608, SEEK_SET) = 1732608
[pid 12935] read(26, "y\2\0\0\1\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\340Z\32\0\0\0\0\0"..., 4096) = 1472
[pid 12935] lseek(26, 1728512, SEEK_SET) = 1728512
[pid 12935] read(26, "setlogin is not implemented and "..., 4096) = 4096
[pid 12935] read(26, "y\2\0\0\1\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\340Z\32\0\0\0\0\0"..., 4096) = 1472
[pid 12935] lseek(26, 0, SEEK_SET) = 0
[pid 12935] read(26, "\177ELF\2\1\1\3\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0P\34\2\0\0\0\0\0"..., 4096) = 4096
[pid 12935] lseek(26, 1728512, SEEK_SET) = 1728512
[pid 12935] read(26, "setlogin is not implemented and "..., 4096) = 4096
[pid 12935] lseek(26, 0, SEEK_SET) = 0
[pid 12935] read(26, "\177ELF\2\1\1\3\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0P\34\2\0\0\0\0\0"..., 4096) = 4096
[pid 12935] --- SIGSEGV (Segmentation fault) @ 0 (0) ---
Process 12935 detached
[pid 12934] +++ killed by SIGSEGV +++
+++ killed by SIGSEGV +++

on strace: strace -p 12928 -f

[fcorriga]

confirmed (3.1 and 3.2). it crashes if you reload the page WHILE the file it's being synched.
it happens thru ftp/sftp too.

[fredemmott]

Fix up internally (D1637887) - should hopefully land soon.

[fredemmott]

Including modified files - same basic issue as #1447