[
https://issues.apache.org/jira/browse/TS-1835?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13888642#comment-13888642
]
Leif Hedstrom commented on TS-1835:
-----------------------------------
Itseems at *least* the struct Resource stuff is pretty much broken (useless),
because of
{code}
Resource *
res_lookup(const char */* path ATS_UNUSED */)
{
return NULL;
}
{code}
It looks like the intent was to make struct Resource a linked list, which seems
like it'd be incredibly expensive at best. This struct Resource is used in
P_IOBuffer.h, i.e.
{code}
#ifdef TRACK_BUFFER_USER
struct Resource;
extern Resource *res_lookup(const char *path);
TS_INLINE void
iobuffer_mem_inc(const char *_loc, int64_t _size_index)
{
...
{code}
Which is never enabled (even if it was, it'd break completely).
> Eliminate ink_resource.{cc,h}
> -----------------------------
>
> Key: TS-1835
> URL: https://issues.apache.org/jira/browse/TS-1835
> Project: Traffic Server
> Issue Type: Bug
> Components: Core
> Reporter: Leif Hedstrom
> Assignee: Leif Hedstrom
> Fix For: 5.0.0
>
>
> I think we should eliminate the ink_resource memory tracking code. We already
> have support for this when linking in with tcmalloc or jemalloc, rolling our
> own seems pointless with these tools being available.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
