Please fill in each of the sections below. Hardware Environment: 2-way SMP P4, IBM Intellistation, Matrox G450 video card Software Environment: RH 7.1, with XFree86-4.0.3-5 Steps to Reproduce: 1.create a .xinitrc in home directory with "exec fvwm2" 2.run "xinit" 3.open an xtern 4.run "xview -onroot pict.jpg" where pict.jpg is a high color picture 5.open another xterm 6.start the fvwm2 WinList module 7.run emacs from the second xterm Actual Results: X crashes/core dumps Expected Results: X emacs should start and X should not crash Additional Information: I found that two core dumps are possible as a result of the same bug. Below is the "bt" output from gdb from each of the cores. #0 0x080aa50f in CopyColormapAndFree () #1 0x080ab971 in QueryColors () #2 0x080abb58 in FreeClientPixels () #3 0x080c41d4 in FreeClientResources () #4 0x080b6946 in CloseDownClient () #5 0x080b0f9b in Dispatch () #6 0x080c20fb in main () #7 0x4007fe5e in __libc_start_main (main=0x80c1b70 , argc=4, . #0 __libc_realloc (oldmem=0xffff, bytes=4) at malloc.c:3243 #1 0x080d904a in Xrealloc () #2 0x080ab613 in FindColor () #3 0x080aaa74 in AllocColor () #4 0x080b52e6 in ProcAllocColor () #5 0x080b0f84 in Dispatch () #6 0x080c20fb in main () #7 0x4007fe5e in __libc_start_main (main=0x80c1b70 , argc=4, . The problem occurs when running X with a 24 bit pixel depth, using overlays, and filling up the colormap. The following patch corrects this problem: . --- xc/programs/Xserver/hw/xfree86/xf8_32bpp/xf86overlay.c Fri May 3 16:35 +++ xc/programs/Xserver/hw/xfree86/xf8_32bpp/xf86overlay.c.fix Fri May 3 16:35 @@ -311,6 +311,7 @@ . pmap->red[pScrn->colorKey].refcnt = AllocPrivate; pmap->red[pScrn->colorKey].fShared = FALSE; + pmap->freeRed--; . color.red = color.blue = color.green = 0; color.pixel = pScrn->colorKey; . This patch is already part of the current XFree86 4.2 cvs tree, but was added after the official 4.2.0 release in January.
Thanks for the patch, I will investigate it and integrate it into our tree.
I've added the patch to rawhide XFree86 4.2.0-53. Thanks again for pointing this fix out to us.