Reclaim oversized-request fds (inline delete) + fix context ctor UAF

Root cause of the earlier 2s0 oversized-delete crashes: the context
constructor initialized p_user=NULL but NOT p_map_params or p_response,
so ~context's 'if (p_map_params) delete p_map_params' / 'if (p_response)
delete p_response' ran on garbage (non-NULL) pointers on the early-delete
paths (EOF-empty, hard read error, oversized) -> a corrupted
std::map<string,string> destructor -> infinite _M_erase recursion ->
SIGSEGV. It was never a libevent issue.

Fix:
- src/sock/context.cpp: initialize p_map_params = NULL and p_response =
  NULL in the constructor (p_user was already NULL). ~context null-guards
  them, so all early-delete paths are now safe.
- src/sock/sock.cpp: the oversized-request branches (headers-incomplete-
  full and body-incomplete-full) now drop the request inline via
  del_event()+delete (exactly like the EOF/hard-error sibling branches)
  instead of leaking the fd/context. This closes the 2s0 fd-exhaustion
  DoS (the deliberate leak tradeoff) with no crash.

Verified: 30 oversized POSTs -> open fd count stays at baseline (7->7,
immediate reclaim, no leak), 0 restarts; normal login + multi-user chat
work. Independent review: APPROVE (preferred the simplification to
inline-delete, applied).

2 files changed, 19 insertions, 10 deletions

diff --git a/ychat/src/sock/context.cpp b/ychat/src/sock/context.cpp
index 0e57fed..11af016 100644
--- a/ychat/src/sock/context.cpp
+++ b/ychat/src/sock/context.cpp
@@ -37,6 +37,8 @@ context::context(sock *p_sock, struct event *p_event, int i_fd)
   this->i_fd = i_fd;
   this->i_buf_len = 0;
   this->c_buf[0] = '\0';
+  this->p_map_params = NULL;
+  this->p_response = NULL;
   this->p_user = NULL;
 }
 
diff --git a/ychat/src/sock/sock.cpp b/ychat/src/sock/sock.cpp
index 9202902..bf733e7 100644
--- a/ychat/src/sock/sock.cpp
+++ b/ychat/src/sock/sock.cpp
@@ -35,6 +35,7 @@
 
 #include "sock.h"
 #include "../tool/tool.h"
+#include "context.h"
 
 using namespace std;
 
@@ -311,13 +312,17 @@ sock::handle_client_read(int i_fd, short event, void *p_arg)
 
     if (i_hdr_end == string::npos)
     {
-      // Headers not fully received yet. Guard against oversized header
-      // floods: if the buffer is already full with no header terminator, the
-      // request can never be completed - drop it (no re-arm) rather than spin.
-      // The fd leaks for that one abusive request (see the body branch below);
-      // preferable to a crash/spin.
+      // Headers not fully received yet. If the buffer is already full with
+      // no header terminator (oversized header flood) the request can never
+      // complete: drop it now (same pattern as the EOF/error branches) rather
+      // than spin. Otherwise wait for the rest.
       if ( p_context->i_buf_len < READSOCK )
         event_add(p_context->p_event, NULL);
+      else
+      {
+        p_context->del_event();
+        delete p_context;
+      }
       return;
     }
 
@@ -332,13 +337,15 @@ sock::handle_client_read(int i_fd, short event, void *p_arg)
       if (i_have < i_content_len)
       {
         // Body still incomplete. If the buffer is already full the body
-        // can never fit (oversized POST): drop the request without re-arming
-        // (re-arming a full buffer would spin, and deleting the context
-        // inside this read callback corrupts libevent). The fd leaks for
-        // that one abusive request — acceptable for a toy chat and far
-        // better than a crash/spin. Otherwise wait for the rest.
+        // can never fit (oversized POST): drop it now (same pattern as the
+        // EOF/error branches) rather than spin. Otherwise wait for the rest.
         if ( p_context->i_buf_len < READSOCK )
           event_add(p_context->p_event, NULL);
+        else
+        {
+          p_context->del_event();
+          delete p_context;
+        }
         return;
       }
     }