Page MenuHomeFreeBSD

Get rid of superfluous "bug marker" code in the bcache
ClosedPublic

Authored by sobomax on Nov 22 2018, 6:16 PM.
Tags
None
Referenced Files
F170721903: D18299.id.diff
Sun, Sep 6, 5:29 AM
Unknown Object (File)
Sat, Sep 5, 8:27 PM
Unknown Object (File)
Sat, Sep 5, 8:27 PM
Unknown Object (File)
Sat, Sep 5, 5:10 PM
Unknown Object (File)
Sat, Sep 5, 1:02 PM
Unknown Object (File)
Sat, Sep 5, 3:42 AM
Unknown Object (File)
Fri, Sep 4, 9:33 AM
Unknown Object (File)
Wed, Sep 2, 10:45 PM
Subscribers

Details

Summary

Read-ahead code introduced in the r313332 contains "the end of cache marker" which I believe superfluous, as cache is allocated on heap and there is already a similar code in the malloc()/free() to detect data corruption and panic:

$ grep USEENDGUARD /usr/src/stand/libsa/*.[ch]
/usr/src/stand/libsa/zalloc_defs.h:#define USEENDGUARD
/usr/src/stand/libsa/zalloc_malloc.c:#ifdef USEENDGUARD
/usr/src/stand/libsa/zalloc_malloc.c:#ifdef USEENDGUARD
/usr/src/stand/libsa/zalloc_malloc.c:#ifdef USEENDGUARD

I believe that code is more "correct" since such incidents should not be taken lightly as often loader logs are not easily available and or often looked upon. Merely issuing a warning when condition is detected is not sufficient since allowing system to continue bootstrap is likely to cause a subsequent kernel crash/misbehaviour that would be really hard to debug and can lead to a severe data corruption as well.

Diff Detail

Repository
rS FreeBSD src repository - subversion
Lint
Lint Skipped
Unit
Tests Skipped
Build Status
Buildable 21161

Event Timeline

Good by me - I did add this for extra safety because the malloc markers are checked on free. And since we havent been bitten yet, I guess we are good:)

This revision is now accepted and ready to land.Nov 22 2018, 6:22 PM
This revision was automatically updated to reflect the committed changes.