Hi again After having reached 2 days of uptime running E17 with 'Idle Cursor' off (i.e. not selected) I can verify that the memory consumption in the X-server has gone back to normal.
top - 13:30:43 up 2 days, 13 min, 3 users, load average: 0.04, 0.17, 0.22 Tasks: 149 total, 1 running, 148 sleeping, 0 stopped, 0 zombie Cpu(s): 3.6%us, 1.9%sy, 5.4%ni, 89.0%id, 0.1%wa, 0.0%hi, 0.0%si, 0.0%st Mem: 1776500k total, 1600716k used, 175784k free, 129676k buffers Swap: 5261248k total, 46600k used, 5214648k free, 917320k cached PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 3371 ingi 20 0 315m 22m 9344 S 2 1.3 48:18.82 enlightenment 2894 root 20 0 584m 145m 5668 S 0 8.4 516:57.03 Xorg I can live without the 'Idle Cursor' candy because of it's cost, but I look forward to seeing the bug fixed in X.org so I can turn it on again. Is/has anyone reported this bug/memleak to X.org ? -- ex animo - Ingi On Sun, Jun 14, 2009 at 11:58 AM, Carsten Haitzler <[email protected]>wrote: > On Sun, 14 Jun 2009 18:53:57 +1000 David Seikel <[email protected]> said: > > > On Fri, 12 Jun 2009 20:23:54 +1000 Carsten Haitzler (The Rasterman) > > <[email protected]> wrote: > > > > > On Thu, 11 Jun 2009 12:03:57 +0200 RocketIII Scientist > > > <[email protected]> said: > > > > > > use xrestop. that reports what x apps use what x resources. you'll > > > find e uses 1 screen sized pixmap for desktop bg and thats pretty > > > much it. as reported already - it looks like x added a leak in mouse > > > pointer changes. the code for animating the pointer (per frame) is: > > > > > > Ecore_X_Cursor cur; > > > > > > cur = ecore_x_cursor_new(p->win, p->pixels, p->w, p->h, > > > p->hot.x, p->hot.y); ecore_x_window_cursor_set(p->win, cur); > > > ecore_x_cursor_free(cur); > > > > > > so literally: > > > 1. create cursor with canvas ARGB pixel data > > > 2. set cursor to window > > > 3. free cursor (ok - as cursor will stay around until the window is > > > deleted due to reference counting). > > > > > > any previously set cursor should be freed if it was set by this as it > > > was already freed and when replaced, will be freed. i bet x has grown > > > a bug in this code as e is probably about the only thing that > > > animates cursor by drawing them its own memory then creating a new > > > cursor for the new frame. > > > > > > you can check the ecore_x calls as to what they do xlib-wise, but > > > they are correct. > > > > So, um, turning off idle cursor will stop that? Or using the X mouse > > cursor? Just so we can A) track it down further for reporting to the > > X.org guys, and B) stop leaking all my damn memory until they get it > > fixed. > > \yes. as you said - or change theme that has no idle cursor or cursor anim > at > all, but it's an xserver leak. needs fixing there. (note - every time e > gets a > click - e emits a signal to the pointer... and the default theme animates a > "glint". a visual key indicating your press was recieved). this will also > leak. > > -- > ------------- Codito, ergo sum - "I code, therefore I am" -------------- > The Rasterman (Carsten Haitzler) [email protected] > > ------------------------------------------------------------------------------ Crystal Reports - New Free Runtime and 30 Day Trial Check out the new simplified licensing option that enables unlimited royalty-free distribution of the report engine for externally facing server and web deployment. http://p.sf.net/sfu/businessobjects _______________________________________________ enlightenment-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/enlightenment-users
