Resent-Date: Mon, 15 Feb 1999 00:57:55 +0100 (MET) To: Jes Sorensen Cc: linux-m68k@lists.linux-m68k.org Subject: Re: Fix for vfork problem (hopefully) References: X-Yow: NATHAN... your PARENTS were in a CARCRASH!! They're VOIDED - They COLLAPSED They had no CHAINSAWS... They had no MONEY MACHINES... They did PILLS in SKIMPY GRASS SKIRTS... Nathan, I EMULATED them... but they were OFF-KEY... From: Andreas Schwab Date: 15 Feb 1999 00:57:50 +0100 In-Reply-To: Jes Sorensen's message of "14 Feb 1999 12:17:12 +0100" Resent-From: linux-m68k@phil.uni-sb.de Unfortunately there is a bug in the patch for pgtable.h, here is the correct one. Andreas. --- linux-2.2/include/asm-m68k/pgtable.h.~1~ Sun Jan 17 22:31:26 1999 +++ linux-2.2/include/asm-m68k/pgtable.h Sat Feb 13 19:07:14 1999 @@ -507,6 +507,7 @@ if (tsk == current) { if (CPU_IS_040_OR_060) __asm__ __volatile__ (".chip 68040\n\t" + "pflushan\n\t" "movec %0,%%urp\n\t" ".chip 68k" : : "r" (tsk->tss.crp[1])); @@ -514,10 +515,22 @@ unsigned long tmp; __asm__ __volatile__ ("movec %%cacr,%0\n\t" "orw #0x0808,%0\n\t" - "movec %0,%%cacr\n\t" - "pmove %1,%%crp\n\t" - : "=d" (tmp) - : "m" (tsk->tss.crp[0])); + "movec %0,%%cacr" + : "=d" (tmp)); + /* For a 030-only kernel, avoid flushing the whole + ATC, we only need to flush the user entries. + The 68851 does this by itself. Avoid a runtime + check here. */ + __asm__ __volatile__ ( +#ifdef CPU_M68030_ONLY + ".chip 68030\n\t" + "pmovefd %0,%%crp\n\t" + ".chip 68k\n\t" + "pflush #0,#4" +#else + "pmove %0,%%crp" +#endif + : : "m" (tsk->tss.crp[0])); } } } -- Andreas Schwab "And now for something schwab@issan.cs.uni-dortmund.de completely different" schwab@gnu.org