CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: riastradh Date: Fri Mar 3 14:32:27 UTC 2023 Modified Files: src/sys/arch/i386/i386: genassym.cf locore.S Log Message: Revert "x86: Add kthread_fpu_enter/exit support, take two -- forgot i386 bits." kthread_fpu_enter/exit changes broke some hardware, unclear why, to investigate before fixing and reapplying these changes. To generate a diff of this commit: cvs rdiff -u -r1.132 -r1.133 src/sys/arch/i386/i386/genassym.cf cvs rdiff -u -r1.193 -r1.194 src/sys/arch/i386/i386/locore.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: riastradh Date: Fri Mar 3 14:32:27 UTC 2023 Modified Files: src/sys/arch/i386/i386: genassym.cf locore.S Log Message: Revert "x86: Add kthread_fpu_enter/exit support, take two -- forgot i386 bits." kthread_fpu_enter/exit changes broke some hardware, unclear why, to investigate before fixing and reapplying these changes. To generate a diff of this commit: cvs rdiff -u -r1.132 -r1.133 src/sys/arch/i386/i386/genassym.cf cvs rdiff -u -r1.193 -r1.194 src/sys/arch/i386/i386/locore.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. Modified files: Index: src/sys/arch/i386/i386/genassym.cf diff -u src/sys/arch/i386/i386/genassym.cf:1.132 src/sys/arch/i386/i386/genassym.cf:1.133 --- src/sys/arch/i386/i386/genassym.cf:1.132 Sat Feb 25 18:35:54 2023 +++ src/sys/arch/i386/i386/genassym.cf Fri Mar 3 14:32:27 2023 @@ -1,4 +1,4 @@ -# $NetBSD: genassym.cf,v 1.132 2023/02/25 18:35:54 riastradh Exp $ +# $NetBSD: genassym.cf,v 1.133 2023/03/03 14:32:27 riastradh Exp $ # # Copyright (c) 1998, 2006, 2007, 2008 The NetBSD Foundation, Inc. @@ -175,7 +175,6 @@ define L_MD_FLAGS offsetof(struct lwp, define L_MD_ASTPENDING offsetof(struct lwp, l_md.md_astpending) define LW_SYSTEM LW_SYSTEM -define LW_SYSTEM_FPU LW_SYSTEM_FPU define MDL_FPU_IN_CPU MDL_FPU_IN_CPU define P_FLAG offsetof(struct proc, p_flag) Index: src/sys/arch/i386/i386/locore.S diff -u src/sys/arch/i386/i386/locore.S:1.193 src/sys/arch/i386/i386/locore.S:1.194 --- src/sys/arch/i386/i386/locore.S:1.193 Wed Mar 1 08:38:50 2023 +++ src/sys/arch/i386/i386/locore.S Fri Mar 3 14:32:27 2023 @@ -1,4 +1,4 @@ -/* $NetBSD: locore.S,v 1.193 2023/03/01 08:38:50 riastradh Exp $ */ +/* $NetBSD: locore.S,v 1.194 2023/03/03 14:32:27 riastradh Exp $ */ /* * Copyright-o-rama! @@ -128,7 +128,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: locore.S,v 1.193 2023/03/01 08:38:50 riastradh Exp $"); +__KERNEL_RCSID(0, "$NetBSD: locore.S,v 1.194 2023/03/03 14:32:27 riastradh Exp $"); #include "opt_copy_symtab.h" #include "opt_ddb.h" @@ -1471,7 +1471,7 @@ ENTRY(cpu_switchto) /* Don't bother with the rest if switching to a system process. */ testl $LW_SYSTEM,L_FLAG(%edi) - jnz .Lswitch_system + jnz switch_return #ifndef XENPV /* Restore thread-private %fs/%gs descriptors. */ @@ -1525,21 +1525,6 @@ switch_return: popl %ebx ret -.Lswitch_system: - /* - * If it has LWP_SYSTEM_FPU set, meaning it's running in - * kthread_fpu_enter/exit, we need to restore the FPU state - * and enable FPU instructions with fpu_handle_deferred. - * - * No need to test MDL_FPU_IN_CPU via HANDLE_DEFERRED_FPU -- - * fpu_switch guarantees it is clear, so we can just call - * fpu_handle_deferred unconditionally. - */ - testl $LW_SYSTEM_FPU,L_FLAG(%edi) - jz switch_return - call _C_LABEL(fpu_handle_deferred) - jmp switch_return - .Lcopy_iobitmap: /* Copy I/O bitmap. */ incl _C_LABEL(pmap_iobmp_evcnt)+EV_COUNT
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: riastradh Date: Sat Feb 25 18:35:54 UTC 2023 Modified Files: src/sys/arch/i386/i386: genassym.cf locore.S Log Message: x86: Add kthread_fpu_enter/exit support, take two -- forgot i386 bits. To generate a diff of this commit: cvs rdiff -u -r1.131 -r1.132 src/sys/arch/i386/i386/genassym.cf cvs rdiff -u -r1.191 -r1.192 src/sys/arch/i386/i386/locore.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. Modified files: Index: src/sys/arch/i386/i386/genassym.cf diff -u src/sys/arch/i386/i386/genassym.cf:1.131 src/sys/arch/i386/i386/genassym.cf:1.132 --- src/sys/arch/i386/i386/genassym.cf:1.131 Tue Dec 27 08:40:40 2022 +++ src/sys/arch/i386/i386/genassym.cf Sat Feb 25 18:35:54 2023 @@ -1,4 +1,4 @@ -# $NetBSD: genassym.cf,v 1.131 2022/12/27 08:40:40 msaitoh Exp $ +# $NetBSD: genassym.cf,v 1.132 2023/02/25 18:35:54 riastradh Exp $ # # Copyright (c) 1998, 2006, 2007, 2008 The NetBSD Foundation, Inc. @@ -175,6 +175,7 @@ define L_MD_FLAGS offsetof(struct lwp, define L_MD_ASTPENDING offsetof(struct lwp, l_md.md_astpending) define LW_SYSTEM LW_SYSTEM +define LW_SYSTEM_FPU LW_SYSTEM_FPU define MDL_FPU_IN_CPU MDL_FPU_IN_CPU define P_FLAG offsetof(struct proc, p_flag) Index: src/sys/arch/i386/i386/locore.S diff -u src/sys/arch/i386/i386/locore.S:1.191 src/sys/arch/i386/i386/locore.S:1.192 --- src/sys/arch/i386/i386/locore.S:1.191 Fri Feb 3 09:21:58 2023 +++ src/sys/arch/i386/i386/locore.S Sat Feb 25 18:35:54 2023 @@ -1,4 +1,4 @@ -/* $NetBSD: locore.S,v 1.191 2023/02/03 09:21:58 mlelstv Exp $ */ +/* $NetBSD: locore.S,v 1.192 2023/02/25 18:35:54 riastradh Exp $ */ /* * Copyright-o-rama! @@ -128,7 +128,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: locore.S,v 1.191 2023/02/03 09:21:58 mlelstv Exp $"); +__KERNEL_RCSID(0, "$NetBSD: locore.S,v 1.192 2023/02/25 18:35:54 riastradh Exp $"); #include "opt_copy_symtab.h" #include "opt_ddb.h" @@ -1447,7 +1447,7 @@ ENTRY(cpu_switchto) /* Don't bother with the rest if switching to a system process. */ testl $LW_SYSTEM,L_FLAG(%edi) - jnz switch_return + jnz .Lswitch_system #ifndef XENPV /* Restore thread-private %fs/%gs descriptors. */ @@ -1501,6 +1501,21 @@ switch_return: popl %ebx ret +.Lswitch_system: + /* + * If it has LWP_SYSTEM_FPU set, meaning it's running in + * kthread_fpu_enter/exit, we need to restore the FPU state + * and enable FPU instructions with fpu_handle_deferred. + * + * No need to test MDL_FPU_IN_CPU via HANDLE_DEFERRED_FPU -- + * fpu_switch guarantees it is clear, so we can just call + * fpu_handle_deferred unconditionally. + */ + testl $LW_SYSTEM_FPU,L_FLAG(%edi) + jz switch_return + call _C_LABEL(fpu_handle_deferred) + jmp switch_return + .Lcopy_iobitmap: /* Copy I/O bitmap. */ incl _C_LABEL(pmap_iobmp_evcnt)+EV_COUNT
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: riastradh Date: Sat Feb 25 18:35:54 UTC 2023 Modified Files: src/sys/arch/i386/i386: genassym.cf locore.S Log Message: x86: Add kthread_fpu_enter/exit support, take two -- forgot i386 bits. To generate a diff of this commit: cvs rdiff -u -r1.131 -r1.132 src/sys/arch/i386/i386/genassym.cf cvs rdiff -u -r1.191 -r1.192 src/sys/arch/i386/i386/locore.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: riastradh Date: Sat Feb 25 13:52:09 UTC 2023 Modified Files: src/sys/arch/i386/i386: bioscall.S Log Message: i386/bioscall: Disable preemption while we load the kernel pmap. Fixes crash on boot in new KASSERT(kpreempt_disabled()) in cpu_load_pmap. We could do KASSERT(kpreempt_disabled() || cold), but I don't know if bioscall is ever reachable after warm boot; this way is safer. XXX Would be nice if bioscall were a C function, not asm -- I don't see any reason why we need it to be asm? To generate a diff of this commit: cvs rdiff -u -r1.10 -r1.11 src/sys/arch/i386/i386/bioscall.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: riastradh Date: Sat Feb 25 13:52:09 UTC 2023 Modified Files: src/sys/arch/i386/i386: bioscall.S Log Message: i386/bioscall: Disable preemption while we load the kernel pmap. Fixes crash on boot in new KASSERT(kpreempt_disabled()) in cpu_load_pmap. We could do KASSERT(kpreempt_disabled() || cold), but I don't know if bioscall is ever reachable after warm boot; this way is safer. XXX Would be nice if bioscall were a C function, not asm -- I don't see any reason why we need it to be asm? To generate a diff of this commit: cvs rdiff -u -r1.10 -r1.11 src/sys/arch/i386/i386/bioscall.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. Modified files: Index: src/sys/arch/i386/i386/bioscall.S diff -u src/sys/arch/i386/i386/bioscall.S:1.10 src/sys/arch/i386/i386/bioscall.S:1.11 --- src/sys/arch/i386/i386/bioscall.S:1.10 Sat Jul 14 14:29:40 2018 +++ src/sys/arch/i386/i386/bioscall.S Sat Feb 25 13:52:09 2023 @@ -1,4 +1,4 @@ -/* $NetBSD: bioscall.S,v 1.10 2018/07/14 14:29:40 maxv Exp $ */ +/* $NetBSD: bioscall.S,v 1.11 2023/02/25 13:52:09 riastradh Exp $ */ /*- * Copyright (c) 1997 The NetBSD Foundation, Inc. @@ -30,7 +30,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: bioscall.S,v 1.10 2018/07/14 14:29:40 maxv Exp $"); +__KERNEL_RCSID(0, "$NetBSD: bioscall.S,v 1.11 2023/02/25 13:52:09 riastradh Exp $"); #include @@ -67,6 +67,8 @@ ENTRY(bioscall) pushl %ebp movl %esp,%ebp /* set up frame ptr */ + call _C_LABEL(kpreempt_disable) + /* install lwp0 pmap */ movl _C_LABEL(kernel_pmap_ptr),%eax pushl %eax @@ -85,5 +87,7 @@ ENTRY(bioscall) call _C_LABEL(cpu_load_pmap) addl $4,%esp + call _C_LABEL(kpreempt_enable) + leave ret
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: mlelstv Date: Fri Feb 3 09:21:58 UTC 2023 Modified Files: src/sys/arch/i386/i386: locore.S Log Message: Make multiboot1 continue with common boot code and not skip CPU type detection. Fixes PR 57017 and PR 56935. To generate a diff of this commit: cvs rdiff -u -r1.190 -r1.191 src/sys/arch/i386/i386/locore.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: mlelstv Date: Fri Feb 3 09:21:58 UTC 2023 Modified Files: src/sys/arch/i386/i386: locore.S Log Message: Make multiboot1 continue with common boot code and not skip CPU type detection. Fixes PR 57017 and PR 56935. To generate a diff of this commit: cvs rdiff -u -r1.190 -r1.191 src/sys/arch/i386/i386/locore.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. Modified files: Index: src/sys/arch/i386/i386/locore.S diff -u src/sys/arch/i386/i386/locore.S:1.190 src/sys/arch/i386/i386/locore.S:1.191 --- src/sys/arch/i386/i386/locore.S:1.190 Wed Sep 7 00:40:18 2022 +++ src/sys/arch/i386/i386/locore.S Fri Feb 3 09:21:58 2023 @@ -1,4 +1,4 @@ -/* $NetBSD: locore.S,v 1.190 2022/09/07 00:40:18 knakahara Exp $ */ +/* $NetBSD: locore.S,v 1.191 2023/02/03 09:21:58 mlelstv Exp $ */ /* * Copyright-o-rama! @@ -128,7 +128,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: locore.S,v 1.190 2022/09/07 00:40:18 knakahara Exp $"); +__KERNEL_RCSID(0, "$NetBSD: locore.S,v 1.191 2023/02/03 09:21:58 mlelstv Exp $"); #include "opt_copy_symtab.h" #include "opt_ddb.h" @@ -415,7 +415,7 @@ multiboot1_loader: pushl %ebx /* Address of Multiboot information */ call _C_LABEL(multiboot1_pre_reloc) addl $4,%esp - jmp 2f + jmp .Lstart_common efi_multiboot2_loader: /*
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: msaitoh Date: Mon Dec 26 08:59:01 UTC 2022 Modified Files: src/sys/arch/i386/i386: genassym.cf Log Message: Remove duplicated IPL_SCHED. No functional change. To generate a diff of this commit: cvs rdiff -u -r1.126 -r1.127 src/sys/arch/i386/i386/genassym.cf Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: msaitoh Date: Mon Dec 26 08:59:01 UTC 2022 Modified Files: src/sys/arch/i386/i386: genassym.cf Log Message: Remove duplicated IPL_SCHED. No functional change. To generate a diff of this commit: cvs rdiff -u -r1.126 -r1.127 src/sys/arch/i386/i386/genassym.cf Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. Modified files: Index: src/sys/arch/i386/i386/genassym.cf diff -u src/sys/arch/i386/i386/genassym.cf:1.126 src/sys/arch/i386/i386/genassym.cf:1.127 --- src/sys/arch/i386/i386/genassym.cf:1.126 Wed Sep 7 00:40:18 2022 +++ src/sys/arch/i386/i386/genassym.cf Mon Dec 26 08:59:01 2022 @@ -1,4 +1,4 @@ -# $NetBSD: genassym.cf,v 1.126 2022/09/07 00:40:18 knakahara Exp $ +# $NetBSD: genassym.cf,v 1.127 2022/12/26 08:59:01 msaitoh Exp $ # # Copyright (c) 1998, 2006, 2007, 2008 The NetBSD Foundation, Inc. @@ -321,7 +321,6 @@ define IS_MASK_COUNT offsetof(struct in define IPL_NONE IPL_NONE define IPL_PREEMPT IPL_PREEMPT define IPL_NET IPL_NET -define IPL_SCHED IPL_SCHED define IPL_CLOCK IPL_CLOCK define IPL_VM IPL_VM define IPL_SCHED IPL_SCHED
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: knakahara Date: Thu Sep 8 06:57:44 UTC 2022 Modified Files: src/sys/arch/i386/i386: lock_stubs.S spl.S Log Message: Fix PR port-i386/57000 (boot failure on qemu). To generate a diff of this commit: cvs rdiff -u -r1.37 -r1.38 src/sys/arch/i386/i386/lock_stubs.S cvs rdiff -u -r1.56 -r1.57 src/sys/arch/i386/i386/spl.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. Modified files: Index: src/sys/arch/i386/i386/lock_stubs.S diff -u src/sys/arch/i386/i386/lock_stubs.S:1.37 src/sys/arch/i386/i386/lock_stubs.S:1.38 --- src/sys/arch/i386/i386/lock_stubs.S:1.37 Wed Sep 7 00:40:18 2022 +++ src/sys/arch/i386/i386/lock_stubs.S Thu Sep 8 06:57:44 2022 @@ -1,4 +1,4 @@ -/* $NetBSD: lock_stubs.S,v 1.37 2022/09/07 00:40:18 knakahara Exp $ */ +/* $NetBSD: lock_stubs.S,v 1.38 2022/09/08 06:57:44 knakahara Exp $ */ /*- * Copyright (c) 2006, 2007, 2008, 2009 The NetBSD Foundation, Inc. @@ -35,7 +35,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: lock_stubs.S,v 1.37 2022/09/07 00:40:18 knakahara Exp $"); +__KERNEL_RCSID(0, "$NetBSD: lock_stubs.S,v 1.38 2022/09/08 06:57:44 knakahara Exp $"); #include "opt_lockdebug.h" @@ -255,10 +255,10 @@ END(mutex_spin_enter) #ifndef XENPV /* - * Release a spin mutex and post a store fence. Must occupy 96 bytes. + * Release a spin mutex and post a store fence. Must occupy 128 bytes. */ ENTRY(mutex_spin_exit) - HOTPATCH(HP_NAME_MUTEX_EXIT, 96) + HOTPATCH(HP_NAME_MUTEX_EXIT, 128) movl 4(%esp), %edx movl CPUVAR(MTX_OLDSPL), %ecx incl CPUVAR(MTX_COUNT) @@ -284,7 +284,7 @@ STRONG_ALIAS(mutex_spin_exit, i686_mutex /* * Patch for i686 CPUs where cli/sti is prohibitively expensive. - * Must be the same size as mutex_spin_exit(), that is, 96 bytes. + * Must be the same size as mutex_spin_exit(), that is, 128 bytes. */ ENTRY(i686_mutex_spin_exit) mov 4(%esp),%edx Index: src/sys/arch/i386/i386/spl.S diff -u src/sys/arch/i386/i386/spl.S:1.56 src/sys/arch/i386/i386/spl.S:1.57 --- src/sys/arch/i386/i386/spl.S:1.56 Wed Sep 7 00:40:18 2022 +++ src/sys/arch/i386/i386/spl.S Thu Sep 8 06:57:44 2022 @@ -1,4 +1,4 @@ -/* $NetBSD: spl.S,v 1.56 2022/09/07 00:40:18 knakahara Exp $ */ +/* $NetBSD: spl.S,v 1.57 2022/09/08 06:57:44 knakahara Exp $ */ /* * Copyright (c) 1998, 2007, 2008, 2020 The NetBSD Foundation, Inc. @@ -30,7 +30,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: spl.S,v 1.56 2022/09/07 00:40:18 knakahara Exp $"); +__KERNEL_RCSID(0, "$NetBSD: spl.S,v 1.57 2022/09/08 06:57:44 knakahara Exp $"); #include "opt_ddb.h" #include "opt_spldebug.h" @@ -72,11 +72,11 @@ END(splraise) * void spllower(int s); * * spllower() for i486 and Pentium. Must be the same size as cx8_spllower(), - * that is, 64 bytes. This must use pushf/cli/popf as it is used early in boot + * that is, 96 bytes. This must use pushf/cli/popf as it is used early in boot * where interrupts are disabled via eflags/IE. */ ENTRY(spllower) - HOTPATCH(HP_NAME_SPLLOWER, 64) + HOTPATCH(HP_NAME_SPLLOWER, 96) #ifdef SPLDEBUG movl 4(%esp),%ecx pushl %ebp @@ -115,7 +115,7 @@ STRONG_ALIAS(spllower, cx8_spllower) * * spllower() optimized for Pentium Pro and later, which have long pipelines * that will be stalled by pushf/cli/popf. Must be the same size as - * spllower(), ie 64 bytes. Does not need to restore eflags/IE as is patched + * spllower(), ie 96 bytes. Does not need to restore eflags/IE as is patched * in once autoconf is underway. * * For cmpxchg8b, edx/ecx are the high words and eax/ebx the low.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: knakahara Date: Thu Sep 8 06:57:44 UTC 2022 Modified Files: src/sys/arch/i386/i386: lock_stubs.S spl.S Log Message: Fix PR port-i386/57000 (boot failure on qemu). To generate a diff of this commit: cvs rdiff -u -r1.37 -r1.38 src/sys/arch/i386/i386/lock_stubs.S cvs rdiff -u -r1.56 -r1.57 src/sys/arch/i386/i386/spl.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: bouyer Date: Tue May 31 18:04:11 UTC 2022 Modified Files: src/sys/arch/i386/i386: locore.S machdep.c Log Message: For some reason on my system, the guest starts with a pending event in the shared structure. On i386, this cause hypervisor_callback to be entered before cpu_info_primary is fully initialised, especially on i386 ci_intrstack is still NULL, which cause a crash when we try to use it. Work around by recycling the boot's tmp stack for this until cpu_attach() is called. To generate a diff of this commit: cvs rdiff -u -r1.187 -r1.188 src/sys/arch/i386/i386/locore.S cvs rdiff -u -r1.834 -r1.835 src/sys/arch/i386/i386/machdep.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. Modified files: Index: src/sys/arch/i386/i386/locore.S diff -u src/sys/arch/i386/i386/locore.S:1.187 src/sys/arch/i386/i386/locore.S:1.188 --- src/sys/arch/i386/i386/locore.S:1.187 Wed Nov 10 15:59:08 2021 +++ src/sys/arch/i386/i386/locore.S Tue May 31 18:04:11 2022 @@ -1,4 +1,4 @@ -/* $NetBSD: locore.S,v 1.187 2021/11/10 15:59:08 msaitoh Exp $ */ +/* $NetBSD: locore.S,v 1.188 2022/05/31 18:04:11 bouyer Exp $ */ /* * Copyright-o-rama! @@ -128,7 +128,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: locore.S,v 1.187 2021/11/10 15:59:08 msaitoh Exp $"); +__KERNEL_RCSID(0, "$NetBSD: locore.S,v 1.188 2022/05/31 18:04:11 bouyer Exp $"); #include "opt_copy_symtab.h" #include "opt_ddb.h" @@ -326,6 +326,7 @@ LABEL(PDPpaddr) .long 0 /* paddr of PDP END(PDPpaddr) /* Space for the temporary stack */ + .globl _C_LABEL(tmpstk) .size tmpstk, tmpstk - . .space 512 tmpstk: Index: src/sys/arch/i386/i386/machdep.c diff -u src/sys/arch/i386/i386/machdep.c:1.834 src/sys/arch/i386/i386/machdep.c:1.835 --- src/sys/arch/i386/i386/machdep.c:1.834 Sun Dec 26 21:33:48 2021 +++ src/sys/arch/i386/i386/machdep.c Tue May 31 18:04:11 2022 @@ -1,4 +1,4 @@ -/* $NetBSD: machdep.c,v 1.834 2021/12/26 21:33:48 riastradh Exp $ */ +/* $NetBSD: machdep.c,v 1.835 2022/05/31 18:04:11 bouyer Exp $ */ /* * Copyright (c) 1996, 1997, 1998, 2000, 2004, 2006, 2008, 2009, 2017 @@ -67,7 +67,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: machdep.c,v 1.834 2021/12/26 21:33:48 riastradh Exp $"); +__KERNEL_RCSID(0, "$NetBSD: machdep.c,v 1.835 2022/05/31 18:04:11 bouyer Exp $"); #include "opt_beep.h" #include "opt_compat_freebsd.h" @@ -1423,6 +1423,8 @@ init386(paddr_t first_avail) #endif #ifdef XENPV + extern int tmpstk; + cpu_info_primary.ci_intrstack = events_default_setup(); #else intr_default_setup();
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: bouyer Date: Tue May 31 18:04:11 UTC 2022 Modified Files: src/sys/arch/i386/i386: locore.S machdep.c Log Message: For some reason on my system, the guest starts with a pending event in the shared structure. On i386, this cause hypervisor_callback to be entered before cpu_info_primary is fully initialised, especially on i386 ci_intrstack is still NULL, which cause a crash when we try to use it. Work around by recycling the boot's tmp stack for this until cpu_attach() is called. To generate a diff of this commit: cvs rdiff -u -r1.187 -r1.188 src/sys/arch/i386/i386/locore.S cvs rdiff -u -r1.834 -r1.835 src/sys/arch/i386/i386/machdep.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src
Committed By: manu
Date: Fri Oct 18 01:19:00 UTC 2019
Modified Files:
src/sys/arch/i386/i386: multiboot.c
Log Message:
Fix multiboot1 kernel symbol load
ELF_Shdr's sh_type field is a value, not a flag field
To generate a diff of this commit:
cvs rdiff -u -r1.24 -r1.25 src/sys/arch/i386/i386/multiboot.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
Modified files:
Index: src/sys/arch/i386/i386/multiboot.c
diff -u src/sys/arch/i386/i386/multiboot.c:1.24 src/sys/arch/i386/i386/multiboot.c:1.25
--- src/sys/arch/i386/i386/multiboot.c:1.24 Thu Nov 8 06:23:48 2018
+++ src/sys/arch/i386/i386/multiboot.c Fri Oct 18 01:19:00 2019
@@ -1,4 +1,4 @@
-/* $NetBSD: multiboot.c,v 1.24 2018/11/08 06:23:48 msaitoh Exp $ */
+/* $NetBSD: multiboot.c,v 1.25 2019/10/18 01:19:00 manu Exp $ */
/*-
* Copyright (c) 2005, 2006 The NetBSD Foundation, Inc.
@@ -30,7 +30,7 @@
*/
#include
-__KERNEL_RCSID(0, "$NetBSD: multiboot.c,v 1.24 2018/11/08 06:23:48 msaitoh Exp $");
+__KERNEL_RCSID(0, "$NetBSD: multiboot.c,v 1.25 2019/10/18 01:19:00 manu Exp $");
#include "opt_multiboot.h"
@@ -302,14 +302,14 @@ copy_syms(struct multiboot_info *mi)
shdrp = &((Elf32_Shdr *)mi->mi_elfshdr_addr)[i];
- if ((shdrp->sh_type & SHT_SYMTAB) &&
+ if ((shdrp->sh_type == SHT_SYMTAB) &&
shdrp->sh_link != SHN_UNDEF) {
Elf32_Shdr *shdrp2;
shdrp2 = &((Elf32_Shdr *)mi->mi_elfshdr_addr)
[shdrp->sh_link];
- if (shdrp2->sh_type & SHT_STRTAB) {
+ if (shdrp2->sh_type == SHT_STRTAB) {
symtabp = shdrp;
strtabp = shdrp2;
}
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: manu Date: Fri Oct 18 01:19:00 UTC 2019 Modified Files: src/sys/arch/i386/i386: multiboot.c Log Message: Fix multiboot1 kernel symbol load ELF_Shdr's sh_type field is a value, not a flag field To generate a diff of this commit: cvs rdiff -u -r1.24 -r1.25 src/sys/arch/i386/i386/multiboot.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: maxv Date: Fri Oct 4 15:28:00 UTC 2019 Modified Files: src/sys/arch/i386/i386: i386_trap.S locore.S spl.S Log Message: Misc reordering, to clarify and reduce the diff against amd64. To generate a diff of this commit: cvs rdiff -u -r1.18 -r1.19 src/sys/arch/i386/i386/i386_trap.S cvs rdiff -u -r1.170 -r1.171 src/sys/arch/i386/i386/locore.S cvs rdiff -u -r1.47 -r1.48 src/sys/arch/i386/i386/spl.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: maxv Date: Fri Oct 4 15:28:00 UTC 2019 Modified Files: src/sys/arch/i386/i386: i386_trap.S locore.S spl.S Log Message: Misc reordering, to clarify and reduce the diff against amd64. To generate a diff of this commit: cvs rdiff -u -r1.18 -r1.19 src/sys/arch/i386/i386/i386_trap.S cvs rdiff -u -r1.170 -r1.171 src/sys/arch/i386/i386/locore.S cvs rdiff -u -r1.47 -r1.48 src/sys/arch/i386/i386/spl.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. Modified files: Index: src/sys/arch/i386/i386/i386_trap.S diff -u src/sys/arch/i386/i386/i386_trap.S:1.18 src/sys/arch/i386/i386/i386_trap.S:1.19 --- src/sys/arch/i386/i386/i386_trap.S:1.18 Mon Feb 11 17:28:52 2019 +++ src/sys/arch/i386/i386/i386_trap.S Fri Oct 4 15:28:00 2019 @@ -1,4 +1,4 @@ -/* $NetBSD: i386_trap.S,v 1.18 2019/02/11 17:28:52 cherry Exp $ */ +/* $NetBSD: i386_trap.S,v 1.19 2019/10/04 15:28:00 maxv Exp $ */ /* * Copyright 2002 (c) Wasabi Systems, Inc. @@ -66,7 +66,7 @@ #if 0 #include -__KERNEL_RCSID(0, "$NetBSD: i386_trap.S,v 1.18 2019/02/11 17:28:52 cherry Exp $"); +__KERNEL_RCSID(0, "$NetBSD: i386_trap.S,v 1.19 2019/10/04 15:28:00 maxv Exp $"); #endif /* @@ -220,7 +220,7 @@ IDTVEC(trap07) pushl %esp call _C_LABEL(fpudna) addl $4,%esp - jmp _C_LABEL(trapreturn) + jmp .Lalltraps_checkusr IDTVEC_END(trap07) IDTVEC(trap08) @@ -282,7 +282,7 @@ IDTVEC(trap0f) #ifdef DIAGNOSTIC movl CPUVAR(ILEVEL),%ebx #endif - jmp _C_LABEL(trapreturn) + jmp .Lalltraps_checkusr IDTVEC_END(trap0f) IDTVEC_END(intrspurious) @@ -302,7 +302,7 @@ IDTVEC(trap10) adcl $0,CPUVAR(NTRAP)+4 call _C_LABEL(fputrap) addl $4,%esp - jmp _C_LABEL(trapreturn) + jmp .Lalltraps_checkusr IDTVEC_END(trap10) IDTVEC(trap11) @@ -394,13 +394,14 @@ IDTVEC_END(tss_trap08) .type _C_LABEL(trap_return_fault_return),@function LABEL(trap_return_fault_return) mov 4(%esp),%esp /* frame for user return */ - jmp _C_LABEL(trapreturn) + jmp .Lalltraps_checkusr END(trap_return_fault_return) /* LINTSTUB: Ignore */ ENTRY(alltraps) INTRENTRY STI(%eax) + calltrap: #ifdef DIAGNOSTIC movl CPUVAR(ILEVEL),%ebx @@ -410,16 +411,18 @@ calltrap: pushl %esp call _C_LABEL(trap) addl $4,%esp -_C_LABEL(trapreturn): .globl trapreturn + +.Lalltraps_checkusr: testb $CHK_UPL,TF_CS(%esp) jnz .Lalltraps_checkast jmp 6f + .Lalltraps_checkast: /* Check for ASTs on exit to user mode. */ CLI(%eax) CHECK_ASTPENDING(%eax) jz 3f -5: CLEAR_ASTPENDING(%eax) + CLEAR_ASTPENDING(%eax) STI(%eax) movl $T_ASTFLT,TF_TRAPNO(%esp) addl $1,CPUVAR(NTRAP) /* statistical info */ @@ -430,6 +433,7 @@ _C_LABEL(trapreturn): .globl trapreturn jmp .Lalltraps_checkast /* re-check ASTs */ 3: CHECK_DEFERRED_SWITCH jnz 9f + #ifdef XEN STIC(%eax) jz 6f @@ -450,15 +454,23 @@ _C_LABEL(trapreturn): .globl trapreturn movl CPUVAR(XSOURCES)(,%eax,4),%eax jmp *IS_RESUME(%eax) 7: movl %ebx,CPUVAR(ILEVEL) /* restore cpl */ - jmp _C_LABEL(trapreturn) -#endif /* XEN */ -#ifndef DIAGNOSTIC -6: INTRFASTEXIT -#else -6: cmpl CPUVAR(ILEVEL),%ebx - jne 3f + jmp .Lalltraps_checkusr +#endif + +6: +#ifdef DIAGNOSTIC + cmpl CPUVAR(ILEVEL),%ebx + jne .Lspl_error +#endif INTRFASTEXIT -3: STI(%eax) + +9: STI(%eax) + call _C_LABEL(pmap_load) + jmp .Lalltraps_checkast /* re-check ASTs */ + +#ifdef DIAGNOSTIC +.Lspl_error: + STI(%eax) pushl $4f call _C_LABEL(panic) addl $4,%esp @@ -467,8 +479,5 @@ _C_LABEL(trapreturn): .globl trapreturn addl $4,%esp jmp .Lalltraps_checkast /* re-check ASTs */ 4: .asciz "SPL NOT LOWERED ON TRAP EXIT\n" -#endif /* DIAGNOSTIC */ -9: STI(%eax) - call _C_LABEL(pmap_load) - jmp .Lalltraps_checkast /* re-check ASTs */ +#endif END(alltraps) Index: src/sys/arch/i386/i386/locore.S diff -u src/sys/arch/i386/i386/locore.S:1.170 src/sys/arch/i386/i386/locore.S:1.171 --- src/sys/arch/i386/i386/locore.S:1.170 Fri Oct 4 11:47:07 2019 +++ src/sys/arch/i386/i386/locore.S Fri Oct 4 15:28:00 2019 @@ -1,4 +1,4 @@ -/* $NetBSD: locore.S,v 1.170 2019/10/04 11:47:07 maxv Exp $ */ +/* $NetBSD: locore.S,v 1.171 2019/10/04 15:28:00 maxv Exp $ */ /* * Copyright-o-rama! @@ -128,7 +128,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: locore.S,v 1.170 2019/10/04 11:47:07 maxv Exp $"); +__KERNEL_RCSID(0, "$NetBSD: locore.S,v 1.171 2019/10/04 15:28:00 maxv Exp $"); #include "opt_copy_symtab.h" #include "opt_ddb.h" @@ -1264,6 +1264,7 @@ IDTVEC(syscall) pushl $T_ASTFLT /* trap # for doing ASTs */ INTRENTRY STI(%eax) + #ifdef DIAGNOSTIC movl CPUVAR(ILEVEL),%ebx testl %ebx,%ebx @@ -1276,6 +1277,7 @@ IDTVEC(syscall) #endif 1: #endif /* DIAGNOSTIC */ + addl $1,CPUVAR(NSYSCALL) /* count it atomically */ adcl $0,CPUVAR(NSYSCALL)+4 /* count it atomically */ movl CPUVAR(CURLWP),%edi @@ -1290,6 +1292,7 @@ IDTVEC(syscall) movl L_MD_ASTPENDING(%edi), %eax orl CPUVAR(WANT_PMAPLOAD), %eax jnz 9f + #ifdef XEN STIC(%eax) jz 14f
CVS commit: src/sys/arch/i386/i386
Module Name:src
Committed By: kamil
Date: Tue Aug 6 02:04:43 UTC 2019
Modified Files:
src/sys/arch/i386/i386: process_machdep.c
Log Message:
Do not dereference user pointer in ptrace_machdep_dorequest()/i386
Always use copyin(9) for reading user memory.
To generate a diff of this commit:
cvs rdiff -u -r1.93 -r1.94 src/sys/arch/i386/i386/process_machdep.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
Modified files:
Index: src/sys/arch/i386/i386/process_machdep.c
diff -u src/sys/arch/i386/i386/process_machdep.c:1.93 src/sys/arch/i386/i386/process_machdep.c:1.94
--- src/sys/arch/i386/i386/process_machdep.c:1.93 Wed Jun 26 12:30:12 2019
+++ src/sys/arch/i386/i386/process_machdep.c Tue Aug 6 02:04:43 2019
@@ -1,4 +1,4 @@
-/* $NetBSD: process_machdep.c,v 1.93 2019/06/26 12:30:12 mgorny Exp $ */
+/* $NetBSD: process_machdep.c,v 1.94 2019/08/06 02:04:43 kamil Exp $ */
/*-
* Copyright (c) 1998, 2000, 2001, 2008 The NetBSD Foundation, Inc.
@@ -75,7 +75,7 @@
*/
#include
-__KERNEL_RCSID(0, "$NetBSD: process_machdep.c,v 1.93 2019/06/26 12:30:12 mgorny Exp $");
+__KERNEL_RCSID(0, "$NetBSD: process_machdep.c,v 1.94 2019/08/06 02:04:43 kamil Exp $");
#include "opt_ptrace.h"
@@ -281,7 +281,7 @@ ptrace_machdep_dorequest(
{
struct uio uio;
struct iovec iov;
- struct iovec *user_iov = (struct iovec*)addr;
+ struct iovec user_iov;
struct vmspace *vm;
int error;
int write = 0;
@@ -319,12 +319,14 @@ ptrace_machdep_dorequest(
/* write = 0 done above. */
if (!process_machdep_validxstate(lt->l_proc))
return EINVAL;
+ if ((error = copyin(addr, _iov, sizeof(user_iov))) != 0)
+ return error;
error = proc_vmspace_getref(l->l_proc, );
if (error) {
return error;
}
- iov.iov_base = user_iov->iov_base;
- iov.iov_len = user_iov->iov_len;
+ iov.iov_base = user_iov.iov_base;
+ iov.iov_len = user_iov.iov_len;
if (iov.iov_len > sizeof(struct xstate))
iov.iov_len = sizeof(struct xstate);
uio.uio_iov =
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: kamil Date: Tue Aug 6 02:04:43 UTC 2019 Modified Files: src/sys/arch/i386/i386: process_machdep.c Log Message: Do not dereference user pointer in ptrace_machdep_dorequest()/i386 Always use copyin(9) for reading user memory. To generate a diff of this commit: cvs rdiff -u -r1.93 -r1.94 src/sys/arch/i386/i386/process_machdep.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: mlelstv Date: Sat Jul 13 17:04:21 UTC 2019 Modified Files: src/sys/arch/i386/i386: trap.c Log Message: Use common code for SMEP/SMAP panic to print more information and allow debugging. To generate a diff of this commit: cvs rdiff -u -r1.301 -r1.302 src/sys/arch/i386/i386/trap.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src
Committed By: mlelstv
Date: Sat Jul 13 17:04:21 UTC 2019
Modified Files:
src/sys/arch/i386/i386: trap.c
Log Message:
Use common code for SMEP/SMAP panic to print more information and
allow debugging.
To generate a diff of this commit:
cvs rdiff -u -r1.301 -r1.302 src/sys/arch/i386/i386/trap.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
Modified files:
Index: src/sys/arch/i386/i386/trap.c
diff -u src/sys/arch/i386/i386/trap.c:1.301 src/sys/arch/i386/i386/trap.c:1.302
--- src/sys/arch/i386/i386/trap.c:1.301 Wed May 29 14:28:37 2019
+++ src/sys/arch/i386/i386/trap.c Sat Jul 13 17:04:21 2019
@@ -1,5 +1,5 @@
-/* $NetBSD: trap.c,v 1.301 2019/05/29 14:28:37 msaitoh Exp $ */
+/* $NetBSD: trap.c,v 1.302 2019/07/13 17:04:21 mlelstv Exp $ */
/*-
* Copyright (c) 1998, 2000, 2005, 2006, 2007, 2008 The NetBSD Foundation, Inc.
@@ -69,7 +69,7 @@
*/
#include
-__KERNEL_RCSID(0, "$NetBSD: trap.c,v 1.301 2019/05/29 14:28:37 msaitoh Exp $");
+__KERNEL_RCSID(0, "$NetBSD: trap.c,v 1.302 2019/07/13 17:04:21 mlelstv Exp $");
#include "opt_ddb.h"
#include "opt_kgdb.h"
@@ -582,17 +582,20 @@ kernelfault:
if (frame->tf_err & PGEX_X) {
/* SMEP might have brought us here */
- if (cr2 > VM_MIN_ADDRESS && cr2 <= VM_MAXUSER_ADDRESS)
-panic("prevented execution of %p (SMEP)",
+ if (cr2 > VM_MIN_ADDRESS && cr2 <= VM_MAXUSER_ADDRESS) {
+printf("prevented execution of %p (SMEP)\n",
(void *)cr2);
+goto we_re_toast;
+ }
}
if ((frame->tf_err & PGEX_P) &&
cr2 < VM_MAXUSER_ADDRESS) {
/* SMAP might have brought us here */
if (onfault_handler(pcb, frame) == NULL) {
-panic("prevented access to %p (SMAP)",
+printf("prevented access to %p (SMAP)\n",
(void *)cr2);
+goto we_re_toast;
}
}
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: msaitoh Date: Wed May 29 14:28:37 UTC 2019 Modified Files: src/sys/arch/i386/i386: trap.c Log Message: Use PRIxREGISTER suggested by Kamil. To generate a diff of this commit: cvs rdiff -u -r1.300 -r1.301 src/sys/arch/i386/i386/trap.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src
Committed By: msaitoh
Date: Wed May 29 14:28:37 UTC 2019
Modified Files:
src/sys/arch/i386/i386: trap.c
Log Message:
Use PRIxREGISTER suggested by Kamil.
To generate a diff of this commit:
cvs rdiff -u -r1.300 -r1.301 src/sys/arch/i386/i386/trap.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
Modified files:
Index: src/sys/arch/i386/i386/trap.c
diff -u src/sys/arch/i386/i386/trap.c:1.300 src/sys/arch/i386/i386/trap.c:1.301
--- src/sys/arch/i386/i386/trap.c:1.300 Sat Apr 6 03:06:25 2019
+++ src/sys/arch/i386/i386/trap.c Wed May 29 14:28:37 2019
@@ -1,5 +1,5 @@
-/* $NetBSD: trap.c,v 1.300 2019/04/06 03:06:25 thorpej Exp $ */
+/* $NetBSD: trap.c,v 1.301 2019/05/29 14:28:37 msaitoh Exp $ */
/*-
* Copyright (c) 1998, 2000, 2005, 2006, 2007, 2008 The NetBSD Foundation, Inc.
@@ -69,7 +69,7 @@
*/
#include
-__KERNEL_RCSID(0, "$NetBSD: trap.c,v 1.300 2019/04/06 03:06:25 thorpej Exp $");
+__KERNEL_RCSID(0, "$NetBSD: trap.c,v 1.301 2019/05/29 14:28:37 msaitoh Exp $");
#include "opt_ddb.h"
#include "opt_kgdb.h"
@@ -467,8 +467,9 @@ kernelfault:
case T_STKFLT|T_USER:
case T_ALIGNFLT|T_USER:
#ifdef TRAP_SIGDEBUG
- printf("pid %d.%d (%s): BUS/SEGV (%#x) at eip %#x addr %#lx\n",
- p->p_pid, l->l_lid, p->p_comm, type, frame->tf_eip, rcr2());
+ printf("pid %d.%d (%s): BUS/SEGV (%#x) at eip %#x addr %#"
+ PRIxREGISTER "\n", p->p_pid, l->l_lid, p->p_comm,
+ type, frame->tf_eip, rcr2());
frame_dump(frame, pcb);
#endif
KSI_INIT_TRAP();
@@ -508,8 +509,9 @@ kernelfault:
case T_PRIVINFLT|T_USER: /* privileged instruction fault */
case T_FPOPFLT|T_USER: /* coprocessor operand fault */
#ifdef TRAP_SIGDEBUG
- printf("pid %d.%d (%s): ILL at eip %#x addr %#lx\n",
- p->p_pid, l->l_lid, p->p_comm, frame->tf_eip, rcr2());
+ printf("pid %d.%d (%s): ILL at eip %#x addr %#"
+ PRIxREGISTER "\n", p->p_pid, l->l_lid, p->p_comm,
+ frame->tf_eip, rcr2());
frame_dump(frame, pcb);
#endif
KSI_INIT_TRAP();
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: maxv Date: Sun May 19 08:49:08 UTC 2019 Modified Files: src/sys/arch/i386/i386: compat_16_machdep.c Log Message: Remove useless call to fpu_sigreset(), buildcontext() already calls it. To generate a diff of this commit: cvs rdiff -u -r1.31 -r1.32 src/sys/arch/i386/i386/compat_16_machdep.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: maxv Date: Sun May 19 08:49:08 UTC 2019 Modified Files: src/sys/arch/i386/i386: compat_16_machdep.c Log Message: Remove useless call to fpu_sigreset(), buildcontext() already calls it. To generate a diff of this commit: cvs rdiff -u -r1.31 -r1.32 src/sys/arch/i386/i386/compat_16_machdep.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. Modified files: Index: src/sys/arch/i386/i386/compat_16_machdep.c diff -u src/sys/arch/i386/i386/compat_16_machdep.c:1.31 src/sys/arch/i386/i386/compat_16_machdep.c:1.32 --- src/sys/arch/i386/i386/compat_16_machdep.c:1.31 Sun May 19 08:46:15 2019 +++ src/sys/arch/i386/i386/compat_16_machdep.c Sun May 19 08:49:08 2019 @@ -1,4 +1,4 @@ -/* $NetBSD: compat_16_machdep.c,v 1.31 2019/05/19 08:46:15 maxv Exp $ */ +/* $NetBSD: compat_16_machdep.c,v 1.32 2019/05/19 08:49:08 maxv Exp $ */ /*- * Copyright (c) 1996, 1997, 1998, 2000 The NetBSD Foundation, Inc. @@ -30,7 +30,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: compat_16_machdep.c,v 1.31 2019/05/19 08:46:15 maxv Exp $"); +__KERNEL_RCSID(0, "$NetBSD: compat_16_machdep.c,v 1.32 2019/05/19 08:49:08 maxv Exp $"); #ifdef _KERNEL_OPT #include "opt_compat_netbsd.h" @@ -232,7 +232,6 @@ sendsig_sigcontext(const ksiginfo_t *ksi /* NOTREACHED */ } - fpu_sigreset(l); buildcontext(l, sel, catcher, fp); /* Remember that we're now on the signal stack. */
Re: CVS commit: src/sys/arch/i386/i386
On Tue, May 31, 2016 at 11:39:08AM +0200, Joerg Sonnenberger wrote: > On Mon, May 30, 2016 at 06:58:52PM -0400, Christos Zoulas wrote: > > Module Name:src > > Committed By: christos > > Date: Mon May 30 22:58:52 UTC 2016 > > > > Modified Files: > > src/sys/arch/i386/i386: cpu_in_cksum.S > > > > Log Message: > > Handle PIC linking for tests > > Except this doesn't work because %ebx is not used as GOT pointer in this > function? Scratch that. Been too long since I wrote i386 assembler... Joerg
Re: CVS commit: src/sys/arch/i386/i386
On Mon, May 30, 2016 at 06:58:52PM -0400, Christos Zoulas wrote: > Module Name: src > Committed By: christos > Date: Mon May 30 22:58:52 UTC 2016 > > Modified Files: > src/sys/arch/i386/i386: cpu_in_cksum.S > > Log Message: > Handle PIC linking for tests Except this doesn't work because %ebx is not used as GOT pointer in this function? Joerg
Re: CVS commit: src/sys/arch/i386/i386
On Thu, Mar 06, 2014 at 12:30:25PM +, NONAKA Kimihiro wrote: Module Name: src Committed By: nonaka Date: Thu Mar 6 12:30:25 UTC 2014 Modified Files: src/sys/arch/i386/i386: cpufunc.S Log Message: fix to pass collect memory address to xrstor. Gah ... :-( FWIW I managed to get gcc 4.8 to optimise some fp loops to use the ymm registers - the xsave/xrstor code seemed to worn an amd64. But I don't have an i386 install on a new enough system. David -- David Laight: da...@l8s.co.uk
Re: CVS commit: src/sys/arch/i386/i386
On Mon, Mar 05, 2012 at 08:01:09AM +0530, Cherry G. Mathew wrote: On 5 March 2012 02:14, Manuel Bouyer bou...@netbsd.org wrote: Module Name: src Committed By: bouyer Date: Sun Mar 4 20:44:17 UTC 2012 Modified Files: src/sys/arch/i386/i386: machdep.c Log Message: Don't try to uvm_page_physload() the tmpgdt page: this always fails because only one physical segment is configured for Xen, and it's probably not worth it to create a second physseg with a single page (uvm has optimisations for the VM_PHYSSEG_MAX == 1 case) ic, so we're potentially leaking 2 pages at boot now We were already leaking 2 pages; as this uvm_page_physload() has never worked ... -- Manuel Bouyer bou...@antioche.eu.org NetBSD: 26 ans d'experience feront toujours la difference --
Re: CVS commit: src/sys/arch/i386/i386
On 5 March 2012 02:14, Manuel Bouyer bou...@netbsd.org wrote: Module Name: src Committed By: bouyer Date: Sun Mar 4 20:44:17 UTC 2012 Modified Files: src/sys/arch/i386/i386: machdep.c Log Message: Don't try to uvm_page_physload() the tmpgdt page: this always fails because only one physical segment is configured for Xen, and it's probably not worth it to create a second physseg with a single page (uvm has optimisations for the VM_PHYSSEG_MAX == 1 case) ic, so we're potentially leaking 2 pages at boot now -- ~Cherry
Re: CVS commit: src/sys/arch/i386/i386
On Fri, Jan 20, 2012 at 12:39:44PM +0100, Matthias Drochner wrote: m...@eterna.com.au said: alloc/free here for whatever is using a lot of memory would be much better than increasing the minimum each LWP requires. Agreed. In the ppbattach case, it should be sufficient to put the devinfo printf into a separate function, so that it doesn't stack up when called recursively. (One just needs to make sure that the compiler doesn't inline it again.) Recursion in the kernel? Isn't that really frowned upon itself? David -- David Laight: da...@l8s.co.uk
Re: CVS commit: src/sys/arch/i386/i386
m...@eterna.com.au said: alloc/free here for whatever is using a lot of memory would be much better than increasing the minimum each LWP requires. Agreed. In the ppbattach case, it should be sufficient to put the devinfo printf into a separate function, so that it doesn't stack up when called recursively. (One just needs to make sure that the compiler doesn't inline it again.) best regards Matthias Forschungszentrum Juelich GmbH 52425 Juelich Sitz der Gesellschaft: Juelich Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498 Vorsitzender des Aufsichtsrats: MinDir Dr. Karl Eugen Huthmacher Geschaeftsfuehrung: Prof. Dr. Achim Bachem (Vorsitzender), Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt, Prof. Dr. Sebastian M. Schmidt Kennen Sie schon unsere app? http://www.fz-juelich.de/app
Re: CVS commit: src/sys/arch/i386/i386
chris...@astron.com said: /.*sub.*,%[er]sp/ very nice 18648 80494b55:stbi_gif_load_from_memory+0xd So this can never have worked on x86, with only 12k stack. best regards Matthias Forschungszentrum Juelich GmbH 52425 Juelich Sitz der Gesellschaft: Juelich Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498 Vorsitzender des Aufsichtsrats: MinDir Dr. Karl Eugen Huthmacher Geschaeftsfuehrung: Prof. Dr. Achim Bachem (Vorsitzender), Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt, Prof. Dr. Sebastian M. Schmidt Kennen Sie schon unsere app? http://www.fz-juelich.de/app
Re: CVS commit: src/sys/arch/i386/i386
dyo...@pobox.com said: increased stack use lead to stack overflow on amd64 with a deep PCI hierarchy Tell me more about this. It was sys/dev/pci/pci.c rev.1.141 which triggered it. Stack use must already have been tight, and the additional device number array was the last straw. The question is now whether it is sufficient in the long run to trim down stack usage (eg devinfo in ppbattach()), or whether the kernel stack needs to be increased. Please revert the reversion. I will add a big, fat comment. It makes more sense if you commit the code together with the comment, and cvs annotate's output would be more sensible too. best regards Matthias Forschungszentrum Juelich GmbH 52425 Juelich Sitz der Gesellschaft: Juelich Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498 Vorsitzender des Aufsichtsrats: MinDir Dr. Karl Eugen Huthmacher Geschaeftsfuehrung: Prof. Dr. Achim Bachem (Vorsitzender), Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt, Prof. Dr. Sebastian M. Schmidt Kennen Sie schon unsere app? http://www.fz-juelich.de/app
re: CVS commit: src/sys/arch/i386/i386
dyo...@pobox.com said: increased stack use lead to stack overflow on amd64 with a deep PCI hierarchy Tell me more about this. It was sys/dev/pci/pci.c rev.1.141 which triggered it. Stack use must already have been tight, and the additional device number array was the last straw. The question is now whether it is sufficient in the long run to trim down stack usage (eg devinfo in ppbattach()), or whether the kernel stack needs to be increased. we should try to decrease kernel stack usage *espcially for* rarely occuring things like autoconfig. alloc/free here for whatever is using a lot of memory would be much better than increasing the minimum each LWP requires. .mrg.
Re: CVS commit: src/sys/arch/i386/i386
On Wed, Jan 18, 2012 at 09:34:39PM +, Matthias Drochner wrote: Module Name: src Committed By: drochner Date: Wed Jan 18 21:34:38 UTC 2012 Modified Files: src/sys/arch/i386/i386: mainbus.c Log Message: revert previous, the assumption all buses 1 and up must be subordinate to pci0 doesn't even hold on i386 -- there are server-class chipsets with multiple primary PCI buses, see arch/x86/pci/pchb.c for examples Matthias, was setting pba_sub = 255 causing material problems for someone, or are you concerned that lossage will eventually occur on certain server-class machines? I ask because, while setting pba_sub that way is far from ideal, it has worked well for me on a variety of machines, and if it has caused anybody any trouble, I don't remember them bringing it up with me. Dave -- David Young dyo...@pobox.comUrbana, IL(217) 721-9981
Re: CVS commit: src/sys/arch/i386/i386
dyo...@pobox.com said: was setting pba_sub = 255 causing material problems for someone, or are you concerned that lossage will eventually occur on certain server-class machines? No, this wasn't causing damage. I was tracking another problem which was incidentally triggered by another of your changes -- increased stack use lead to stack overflow on amd64 with a deep PCI hierarchy (a MicroTCA system). Just stumbled over this. But it is obviously wrong. If you think it should be in the public tree there should be at least a big fat comment telling that it is for testing purposes and nothing uses it really and it will be removed before it can cause damage. best regards Matthias Forschungszentrum Juelich GmbH 52425 Juelich Sitz der Gesellschaft: Juelich Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498 Vorsitzender des Aufsichtsrats: MinDir Dr. Karl Eugen Huthmacher Geschaeftsfuehrung: Prof. Dr. Achim Bachem (Vorsitzender), Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt, Prof. Dr. Sebastian M. Schmidt Kennen Sie schon unsere app? http://www.fz-juelich.de/app
Re: CVS commit: src/sys/arch/i386/i386
On Wed, Jan 18, 2012 at 11:38:39PM +0100, Matthias Drochner wrote: dyo...@pobox.com said: was setting pba_sub = 255 causing material problems for someone, or are you concerned that lossage will eventually occur on certain server-class machines? No, this wasn't causing damage. I was tracking another problem which was incidentally triggered by another of your changes -- increased stack use lead to stack overflow on amd64 with a deep PCI hierarchy (a MicroTCA system). Tell me more about this. Just stumbled over this. But it is obviously wrong. If you think it should be in the public tree there should be at least a big fat comment telling that it is for testing purposes and nothing uses it really and it will be removed before it can cause damage. It does belong in the public tree. Please revert the reversion. I will add a big, fat comment. Dave -- David Young dyo...@pobox.comUrbana, IL(217) 721-9981
Re: CVS commit: src/sys/arch/i386/i386
hi, On Oct 31, 2011, at 1:42 PM, YAMAMOTO Takashi wrote: Module Name: src Committed By:yamt Date:Mon Oct 31 12:42:53 UTC 2011 Modified Files: src/sys/arch/i386/i386: dumpsys.c Log Message: dumpsys_seg: don't overwrite the previous mapping With this change in place core dumps from ddb (reboot 104) no longer work on MP machines. Before pmap_tlb_shootnow() always returned on the `tp-tp_count == 0' check. Now it goes into the `remote' case and hangs hard trying to reach other CPUs. thanks, i've reverted the changes for now. YAMAMOTO Takashi -- Juergen Hannken-Illjes - hann...@eis.cs.tu-bs.de - TU Braunschweig (Germany)
Re: CVS commit: src/sys/arch/i386/i386
J. Hannken-Illjes hann...@eis.cs.tu-bs.de wrote: Module Name:src Committed By: yamt Date: Mon Oct 31 12:42:53 UTC 2011 Modified Files: src/sys/arch/i386/i386: dumpsys.c Log Message: dumpsys_seg: don't overwrite the previous mapping With this change in place core dumps from ddb (reboot 104) no longer work on MP machines. Before pmap_tlb_shootnow() always returned on the `tp-tp_count == 0' check. Now it goes into the `remote' case and hangs hard trying to reach other CPUs. We could check for panicstr. Not sure if that is desirable, though. -- Mindaugas
Re: CVS commit: src/sys/arch/i386/i386
On Oct 31, 2011, at 1:42 PM, YAMAMOTO Takashi wrote: Module Name: src Committed By: yamt Date: Mon Oct 31 12:42:53 UTC 2011 Modified Files: src/sys/arch/i386/i386: dumpsys.c Log Message: dumpsys_seg: don't overwrite the previous mapping With this change in place core dumps from ddb (reboot 104) no longer work on MP machines. Before pmap_tlb_shootnow() always returned on the `tp-tp_count == 0' check. Now it goes into the `remote' case and hangs hard trying to reach other CPUs. -- Juergen Hannken-Illjes - hann...@eis.cs.tu-bs.de - TU Braunschweig (Germany)
Re: CVS commit: src/sys/arch/i386/i386
On 01/12/2011 05:09 AM, enami tsugutomo wrote: =?UTF-8?B?R3LDqWdvaXJlIFN1dHJl?=gsu...@netbsd.org writes: Assume for instance that the boot-loader left us with: +--+ ++ +--+ | string table | | kernel | | symbol table | +--+ ++ +--+ The new addresses computed by lines 338-359 (here, it's really lines 344-345) will move the tables so that they end up as: ++--+--+ | kernel | symbol table | string table | ++--+--+ If this ascii art is correct, memmove should be used instead of memcpy. Agreed. I should have mentioned it in the commit message. Here, I just wanted to restore previous (working) behavior. Note that the figure is just an example: afaik, no assumption can be made regarding the placement of the tables. Also, if initial order is kernel, string table, symbol table and a gap between kernel and string table is smaller than symbol table, copying it may overwrite the string table. Is there any guarantee that the gap is large enough? The code (that was already there) takes care of this by examining all possible cases. If initial order is kernel, string table, symbol table, then the order remains the same. Grégoire
Re: CVS commit: src/sys/arch/i386/i386
=?UTF-8?B?R3LDqWdvaXJlIFN1dHJl?= gsu...@netbsd.org writes: Assume for instance that the boot-loader left us with: +--+ ++ +--+ | string table | | kernel | | symbol table | +--+ ++ +--+ The new addresses computed by lines 338-359 (here, it's really lines 344-345) will move the tables so that they end up as: ++--+--+ | kernel | symbol table | string table | ++--+--+ If this ascii art is correct, memmove should be used instead of memcpy. Also, if initial order is kernel, string table, symbol table and a gap between kernel and string table is smaller than symbol table, copying it may overwrite the string table. Is there any guarantee that the gap is large enough? enami.
Re: CVS commit: src/sys/arch/i386/i386
On Sun, Oct 31, 2010 at 04:51:19AM +, YAMAMOTO Takashi wrote: Module Name: src Committed By: yamt Date: Sun Oct 31 04:51:19 UTC 2010 Modified Files: src/sys/arch/i386/i386: vector.S Log Message: keep interrupts disabled in NMI handler. the patch provided by IRINO yoshiaki in PR/43007. Good catch! Dave -- David Young OJC Technologies dyo...@ojctech.com Urbana, IL * (217) 278-3933
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: jym Date: Mon Feb 22 23:52:17 UTC 2010 Modified Files: src/sys/arch/i386/i386: genassym.cf Log Message: Remove a duplicate entry for PTE_BASE. While here, introduce PDP_SIZE and PDE_SIZE. They will be used by asm code to make it more generic (PAE related). To generate a diff of this commit: cvs rdiff -u -r1.84 -r1.85 src/sys/arch/i386/i386/genassym.cf Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: jym Date: Mon Feb 22 23:52:17 UTC 2010 Modified Files: src/sys/arch/i386/i386: genassym.cf Log Message: Remove a duplicate entry for PTE_BASE. While here, introduce PDP_SIZE and PDE_SIZE. They will be used by asm code to make it more generic (PAE related). To generate a diff of this commit: cvs rdiff -u -r1.84 -r1.85 src/sys/arch/i386/i386/genassym.cf Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. Modified files: Index: src/sys/arch/i386/i386/genassym.cf diff -u src/sys/arch/i386/i386/genassym.cf:1.84 src/sys/arch/i386/i386/genassym.cf:1.85 --- src/sys/arch/i386/i386/genassym.cf:1.84 Fri Nov 27 03:23:10 2009 +++ src/sys/arch/i386/i386/genassym.cf Mon Feb 22 23:52:17 2010 @@ -1,4 +1,4 @@ -# $NetBSD: genassym.cf,v 1.84 2009/11/27 03:23:10 rmind Exp $ +# $NetBSD: genassym.cf,v 1.85 2010/02/22 23:52:17 jym Exp $ # # Copyright (c) 1998, 2006, 2007, 2008 The NetBSD Foundation, Inc. @@ -150,7 +150,6 @@ define LSRUN LSRUN define LSONPROC LSONPROC -define PTE_BASE PTE_BASE define UPAGES UPAGES define USPACE USPACE define KSTACK_SIZE KSTACK_SIZE @@ -169,6 +168,9 @@ define PDIR_SLOT_PTE PDIR_SLOT_PTE define PTE_BASE PTE_BASE +define PDP_SIZE PDP_SIZE +define PDE_SIZE sizeof(pd_entry_t) + define IOMAPSIZE IOMAPSIZE define VM_MAXUSER_ADDRESS (int)VM_MAXUSER_ADDRESS
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: darran Date: Mon Feb 22 06:42:15 UTC 2010 Modified Files: src/sys/arch/i386/i386: trap.c vector.S Log Message: DTrace: missed the opt_dtrace.h include for these. To generate a diff of this commit: cvs rdiff -u -r1.254 -r1.255 src/sys/arch/i386/i386/trap.c cvs rdiff -u -r1.52 -r1.53 src/sys/arch/i386/i386/vector.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
Re: CVS commit: src/sys/arch/i386/i386
da...@l8s.co.uk said: I presume the panic was in the 'return to user' sequence? Yes, in the iret. Does it now SIGSEGV the program? Yes, which is a bit strange because this happened before even the first instruction was fetched, and no user-supplied data are involved (as opposed to setmcontext). It was entirely the kernel's fault, so a panic would be only fair. (putting in the kernel %cs might leave you running user code with kernel privelege) This couldn't happen here because the CPL=3 is set together with the CS descriptor, so one should get a GPF on any attempt to use a kernel descriptor. best regards Matthias Forschungszentrum Juelich GmbH 52425 Juelich Sitz der Gesellschaft: Juelich Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498 Vorsitzende des Aufsichtsrats: MinDir'in Baerbel Brumme-Bothe Geschaeftsfuehrung: Prof. Dr. Achim Bachem (Vorsitzender), Dr. Ulrich Krafft (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt, Prof. Dr. Sebastian M. Schmidt
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: drochner Date: Sun Feb 14 11:09:54 UTC 2010 Modified Files: src/sys/arch/i386/i386: ibcs2_machdep.c svr4_machdep.c Log Message: fix confused CS selector, fixes the panic reported by Mark Davis per PR port-i386/42787 (the panic happens due to a GPF when a privileged descriptor is tried to be loaded with the UPL bit set) The original bug is very old (pre-2.0, i386/svr4_machdep.c rev. 1.69), but it was relatively harmless until the order of GDT entries was shuffled (pre-5.0, i386/segments.h rev. 1.42). Before, it caused a userlevel data selector to be used for CS which broke the emulation (likely the reason of PR port-i386/32424). The shuffle made that a privileged selector was used, causing the GPF. (recent -current doesn't panic on that GPF which seems to be a side effect of another change) To generate a diff of this commit: cvs rdiff -u -r1.39 -r1.40 src/sys/arch/i386/i386/ibcs2_machdep.c cvs rdiff -u -r1.95 -r1.96 src/sys/arch/i386/i386/svr4_machdep.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: jym Date: Tue Feb 9 23:09:47 UTC 2010 Modified Files: src/sys/arch/i386/i386: mptramp.S Log Message: Use CR0_PE (enable protected mode) instead of hardcoding constant. To generate a diff of this commit: cvs rdiff -u -r1.19 -r1.20 src/sys/arch/i386/i386/mptramp.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: dsl Date: Sun Jan 17 22:21:18 UTC 2010 Modified Files: src/sys/arch/i386/i386: trap.c vector.S Log Message: Fix 'fault on load of %gs during retirn to userspace' to look for the correct instruction bytes. Take the 'fault on load segment register' through the same path as 'fault on iret' so we don't have to fixup the broken stackframe that contains a mix of user and kernel registers, Update comments about how the faults during return to userspace are processed. Setting an invalid %gs in the saved context of a signal handler causes a SIGSEGV handler to be entered with what look like valid registers. To generate a diff of this commit: cvs rdiff -u -r1.252 -r1.253 src/sys/arch/i386/i386/trap.c cvs rdiff -u -r1.50 -r1.51 src/sys/arch/i386/i386/vector.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: dsl Date: Sun Jan 10 15:21:36 UTC 2010 Modified Files: src/sys/arch/i386/i386: trap.c vector.S Log Message: If we fault on the 'iret' during return to userpace (eg if %eip is outside the bounds of %cs) then hack the stack to contain a normal fault frame for the signal setup code (etc). Previously the code assumed that the original user trap frame was still present - at it is for faults when loading the segment registers. To generate a diff of this commit: cvs rdiff -u -r1.250 -r1.251 src/sys/arch/i386/i386/trap.c cvs rdiff -u -r1.49 -r1.50 src/sys/arch/i386/i386/vector.S Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: dsl Date: Sun Jan 10 15:37:36 UTC 2010 Modified Files: src/sys/arch/i386/i386: trap.c Log Message: If we fault on the iret during return to userspace, see if we need to do a lazy update of %cs to make the stack executable. If a change is made, just retry the failing sequence. Signal handlers as gcc nested local functions now work! To generate a diff of this commit: cvs rdiff -u -r1.251 -r1.252 src/sys/arch/i386/i386/trap.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/sys/arch/i386/i386
Module Name:src Committed By: christos Date: Thu Dec 31 16:04:32 UTC 2009 Modified Files: src/sys/arch/i386/i386: apmbios.c Log Message: Fix compilation issue with time_t To generate a diff of this commit: cvs rdiff -u -r1.15 -r1.16 src/sys/arch/i386/i386/apmbios.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
Re: CVS commit: src/sys/arch/i386/i386
hi, Hello, David Laight wrote: Module Name: src Committed By:dsl Date:Sat Nov 21 19:59:51 UTC 2009 Modified Files: src/sys/arch/i386/i386: db_trace.c Log Message: Don't call lwp_getpcb() when not KERNEL. Thanks for working this around, however it is not a right fix. With this patch, build.sh release passes: http://www.netbsd.org/~rmind/crash_pcb.diff Requesting http://www.netbsd.org/~rmind/crash_pcb.diff ftp: Error retrieving file `404 Not Found' YAMAMOTO Takashi It should have been enough to amend the argument of routine exported to userland. Do not you agree? -- Mindaugas
Re: CVS commit: src/sys/arch/i386/i386
David Laight da...@l8s.co.uk wrote: Thanks for working this around, however it is not a right fix. With this patch, build.sh release passes: http://www.netbsd.org/~rmind/crash_pcb.diff It should have been enough to amend the argument of routine exported to userland. Do not you agree? ... No - you shouldn't be making any names visible to normal user programs. We pollute user namespace far too much as it is. Can you please elaborate on this? Normal user application do not and should not include sys/lwp.h, only KVM grovellers (ps, libkvm) do that. There are few utilities which do that via sys/proc.h, but lwp.h already exports stuff like USER_TO_UAREA/UAREA_TO_USER. Addition of lwp_getpcb(), which is supposed to replace them, does not change the status quo. Do not you agree? Having said that, it might be useful to abstract them using on _LWP_PRIVATE definition or similar. However, it is unrelated to this change. Moreover, your revert changed the original definition of lwp_getpcb(). The 'fix' I did was a pragmatic one to get the build working. Programs like crash (and gdb) need to be able to define something in order to get these extra structures (etc) visible. -- Mindaugas
Re: CVS commit: src/sys/arch/i386/i386
Hello, David Laight wrote: Module Name: src Committed By: dsl Date: Sat Nov 21 19:59:51 UTC 2009 Modified Files: src/sys/arch/i386/i386: db_trace.c Log Message: Don't call lwp_getpcb() when not KERNEL. Thanks for working this around, however it is not a right fix. With this patch, build.sh release passes: http://www.netbsd.org/~rmind/crash_pcb.diff It should have been enough to amend the argument of routine exported to userland. Do not you agree? -- Mindaugas
Re: CVS commit: src/sys/arch/i386/i386
In article 20091121223029.1e0d263b...@mail.netbsd.org, Mindaugas Rasiukevicius rm...@netbsd.org wrote: Hello, David Laight wrote: Module Name: src Committed By:dsl Date:Sat Nov 21 19:59:51 UTC 2009 Modified Files: src/sys/arch/i386/i386: db_trace.c Log Message: Don't call lwp_getpcb() when not KERNEL. Thanks for working this around, however it is not a right fix. With this patch, build.sh release passes: http://www.netbsd.org/~rmind/crash_pcb.diff It should have been enough to amend the argument of routine exported to userland. Do not you agree? I do. christos
Re: CVS commit: src/sys/arch/i386/i386
On Sat, May 09, 2009 at 06:58:46AM +, Andrew Doran wrote: xen isn't as vulnerable to the LDT/segreg problem as native x86 because it's not MP and doesn't do kernel preemption. For the time being I guess it would suffice to #ifdef the 'cli'. That's not enough to make the binary run: even with the cli commented out the test binary segfaults. There may be something else wrong, I'll try to look further. Any way to use gdb to see what it's doing ? The NetBSD/i386 doesn't want to load this executable ... Can you ktrace it? At least we then can see if it's hitting the syscall path. here's what I get: 11703 1 ktrace EMUL netbsd 11703 1 ktrace RET ktrace 0 11703 1 ktrace CALL execve(0xbf7ffc27,0xbf7feb68,0xbf7feb70) 11703 1 ktrace NAMI ./architextIndex 11703 1 architextIndex EMUL netbsd 11703 1 architextIndex RET syscall JUSTRETURN 11703 1 architextIndex PSIG SIGSEGV SIG_DFL: code=SEGV_ACCERR, addr=0xacb94, trap=4) 11703 1 architextIndex NAMI architextIndex.core -- Manuel Bouyer, LIP6, Universite Paris VI. manuel.bou...@lip6.fr NetBSD: 26 ans d'experience feront toujours la difference --
Re: CVS commit: src/sys/arch/i386/i386
On Tue, May 05, 2009 at 09:20:02PM +, Andrew Doran wrote: but the test binary from the PR segfaults: truc# kdump |less 34 0 ktrace EMUL netbsd 34 0 ktrace RET ktrace 0 34 0 ktrace CALL execve(0xbf7ffc02,0xbf7feb3c,0xbf7feb44) 34 0 ktrace NAMI ./architextIndex 34 0 architextIndex EMUL netbsd 34 0 architextIndex RET syscall JUSTRETURN 34 0 architextIndex PSIG SIGSEGV SIG_DFL: code=SEGV_ACCERR, addr=0xacb 94, trap=4) 34 0 architextIndex NAMI architextIndex.core On Xen CLI(%eax) expands to: movlCPUVAR(VCPU),%eax ; movb $1,EVTCHN_UPCALL_MASK(%eax) At this point the segment registers won't be set up. And %eax contains the syscall number. I guess this is a problem. Is there a way to account for this somewhere ? It is difficult to avoid the LDT/segreg problems without having interrupts disabled instantly on entry. Maybe we could add really ugly logic to compensate for it in trap() since oosyscall is the only place where we enter with interupts on (I don't know how interrupts/traps are set up on xen currently). xen isn't as vulnerable to the LDT/segreg problem as native x86 because it's not MP and doesn't do kernel preemption. For the time being I guess it would suffice to #ifdef the 'cli'. That's not enough to make the binary run: even with the cli commented out the test binary segfaults. There may be something else wrong, I'll try to look further. Any way to use gdb to see what it's doing ? The NetBSD/i386 doesn't want to load this executable ... -- Manuel Bouyer bou...@antioche.eu.org NetBSD: 26 ans d'experience feront toujours la difference --
Re: CVS commit: src/sys/arch/i386/i386
On Mon, May 04, 2009 at 11:47:29AM +, Andrew Doran wrote: Module Name: src Committed By: ad Date: Mon May 4 11:47:29 UTC 2009 Modified Files: src/sys/arch/i386/i386: locore.S Log Message: PR kern/41342: BSDi binaries cause panic XXX Manuel, please have a look as I am not sure what to do for XEN here! I tried the following code instead: IDTVEC(osyscall) CLI(%eax) # must be first instruction pushfl # set eflags in trap frame popl8(%esp) orl $PSL_I,(%esp) # re-enable ints on return to user pushl $7 # size of instruction for restart jmp syscall1 but the test binary from the PR segfaults: truc# kdump |less 34 0 ktrace EMUL netbsd 34 0 ktrace RET ktrace 0 34 0 ktrace CALL execve(0xbf7ffc02,0xbf7feb3c,0xbf7feb44) 34 0 ktrace NAMI ./architextIndex 34 0 architextIndex EMUL netbsd 34 0 architextIndex RET syscall JUSTRETURN 34 0 architextIndex PSIG SIGSEGV SIG_DFL: code=SEGV_ACCERR, addr=0xacb 94, trap=4) 34 0 architextIndex NAMI architextIndex.core On Xen CLI(%eax) expands to: movlCPUVAR(VCPU),%eax ; movb $1,EVTCHN_UPCALL_MASK(%eax) I guess this is a problem. Is there a way to account for this somewhere ? -- Manuel Bouyer, LIP6, Universite Paris VI. manuel.bou...@lip6.fr NetBSD: 26 ans d'experience feront toujours la difference --
Re: CVS commit: src/sys/arch/i386/i386
dyo...@netbsd.org wrote:
Module Name: src
Committed By: dyoung
Date: Mon Mar 30 17:48:22 UTC 2009
Modified Files:
src/sys/arch/i386/i386: vm_machdep.c
Log Message:
Cosmetic: reduce differences from amd64/amd64/vm_machdep.c by deleting
an empty line.
src/share/misc/style says to insert it:
---
static int
dirinfo(const char *p, struct stat *sb, struct dirent *de, struct statfs *sf,
int *rargc, char **rargv[])
{ /* Insert an empty line if the function has no local variables. */
/*
* In system libraries, catch obviously invalid function arguments
* using _DIAGASSERT(3).
*/
_DIAGASSERT(p != NULL);
---
Izumi Tsutsui
