nhyatt/putty-source
1
0
Fork 0
mirror of https://git.tartarus.org/simon/putty.git synced 2025-01-10 01:48:00 +00:00
Code Issues Projects Releases Wiki Activity

Reinstate RDB's pending_netevent mechanism, which was removed in

Browse Source 
r4906 in the process of adding the new timing code. It seems to have
been what was previously preventing spew-lockup, and still seems to
prevent it now I've put it back in.

[originally from svn r7071]
[r4906 == 7ecf13564a]
[this svn revision also touched putty-wishlist]
This commit is contained in:
Simon Tatham 2007-01-07 12:40:21 +00:00
parent c1a9dbef13
commit 57a763b0ec
1 changed files with 24 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

29
windows/window.c
View File

@ -100,7 +100,10 @@ static int offset_width, offset_height;
static int was_zoomed = 0;
static int prev_rows, prev_cols;
static void enact_netevent(WPARAM, LPARAM);
static int pending_netevent = 0;
static WPARAM pend_netevent_wParam = 0;
static LPARAM pend_netevent_lParam = 0;
static void enact_pending_netevent(void);
static void flash_window(int mode);
static void sys_cursor_update(void);
static int is_shift_pressed(void);
@ -808,7 +811,6 @@ int WINAPI WinMain(HINSTANCE inst, HINSTANCE prev, LPSTR cmdline, int show)
sfree(handles);
if (must_close_session)
close_session();
continue;
}
sfree(handles);
@ -832,6 +834,9 @@ int WINAPI WinMain(HINSTANCE inst, HINSTANCE prev, LPSTR cmdline, int show)
/* The messages seem unreliable; especially if we're being tricky */
term_set_focus(term, GetForegroundWindow() == hwnd);
if (pending_netevent)
enact_pending_netevent();
net_pending_errors();
}
@ -1074,7 +1079,7 @@ void cmdline_error(char *fmt, ...)
/*
* Actually do the job requested by a WM_NETEVENT
*/
static void enact_netevent(WPARAM wParam, LPARAM lParam)
static void enact_pending_netevent(void)
{
static int reentering = 0;
extern int select_result(WPARAM, LPARAM);
@ -1082,8 +1087,10 @@ static void enact_netevent(WPARAM wParam, LPARAM lParam)
if (reentering)
return; /* don't unpend the pending */
pending_netevent = FALSE;
reentering = 1;
select_result(wParam, lParam);
select_result(pend_netevent_wParam, pend_netevent_lParam);
reentering = 0;
}
@ -2565,7 +2572,19 @@ static LRESULT CALLBACK WndProc(HWND hwnd, UINT message,
}
return 0;
case WM_NETEVENT:
enact_netevent(wParam, lParam);
/* Notice we can get multiple netevents, FD_READ, FD_WRITE etc
* but the only one that's likely to try to overload us is FD_READ.
* This means buffering just one is fine.
*/
if (pending_netevent)
enact_pending_netevent();
pending_netevent = TRUE;
pend_netevent_wParam = wParam;
pend_netevent_lParam = lParam;
if (WSAGETSELECTEVENT(lParam) != FD_READ)
enact_pending_netevent();
net_pending_errors();
return 0;
case WM_SETFOCUS: