Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1759067AbZDGGhv (ORCPT ); Tue, 7 Apr 2009 02:37:51 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1758733AbZDGGgv (ORCPT ); Tue, 7 Apr 2009 02:36:51 -0400 Received: from e28smtp01.in.ibm.com ([59.145.155.1]:56057 "EHLO e28smtp01.in.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758180AbZDGGgt (ORCPT ); Tue, 7 Apr 2009 02:36:49 -0400 Date: Tue, 7 Apr 2009 12:06:16 +0530 From: "K.Prasad" To: Alan Stern Cc: Ingo Molnar , Linux Kernel Mailing List , Andrew Morton , Benjamin Herrenschmidt , Frederic Weisbecker , maneesh@linux.vnet.ibm.com, Roland McGrath , Steven Rostedt , "K.Prasad" Subject: [Patch 06/11] Use the new wrapper routines to access debug registers in process/thread code Message-ID: <20090407063616.GG17461@in.ibm.com> References: <20090407063058.301701787@prasadkr_t60p.in.ibm.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline; filename=modify_process_related_06 User-Agent: Mutt/1.5.18 (2008-05-17) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 8026 Lines: 240 From: Alan Stern This patch enables the use of abstract debug registers in process-handling routines. [K.Prasad: Split-out from the bigger patch and minor changes following re-basing] Signed-off-by: K.Prasad Signed-off-by: Alan Stern --- arch/x86/include/asm/processor.h | 4 --- arch/x86/kernel/process.c | 23 ++++------------- arch/x86/kernel/process_32.c | 31 +++++++++++++++++++++++ arch/x86/kernel/process_64.c | 33 +++++++++++++++++++++++++ 4 files changed, 70 insertions(+), 21 deletions(-) Index: arch/x86/kernel/process.c =================================================================== --- arch/x86/kernel/process.c.orig 2009-04-01 20:53:43.000000000 +0530 +++ arch/x86/kernel/process.c 2009-04-01 20:54:39.000000000 +0530 @@ -14,6 +14,8 @@ #include #include #include +#include +#include unsigned long idle_halt; EXPORT_SYMBOL(idle_halt); @@ -83,6 +85,8 @@ put_cpu(); kfree(bp); } + if (unlikely(test_tsk_thread_flag(me, TIF_DEBUG))) + flush_thread_hw_breakpoint(me); ds_exit_thread(current); } @@ -103,14 +107,9 @@ } #endif - clear_tsk_thread_flag(tsk, TIF_DEBUG); + if (unlikely(test_tsk_thread_flag(tsk, TIF_DEBUG))) + flush_thread_hw_breakpoint(tsk); - tsk->thread.debugreg0 = 0; - tsk->thread.debugreg1 = 0; - tsk->thread.debugreg2 = 0; - tsk->thread.debugreg3 = 0; - tsk->thread.debugreg6 = 0; - tsk->thread.debugreg7 = 0; memset(tsk->thread.tls_array, 0, sizeof(tsk->thread.tls_array)); /* * Forget coprocessor state.. @@ -192,16 +191,6 @@ else if (next->debugctlmsr != prev->debugctlmsr) update_debugctlmsr(next->debugctlmsr); - if (test_tsk_thread_flag(next_p, TIF_DEBUG)) { - set_debugreg(next->debugreg0, 0); - set_debugreg(next->debugreg1, 1); - set_debugreg(next->debugreg2, 2); - set_debugreg(next->debugreg3, 3); - /* no 4 and 5 */ - set_debugreg(next->debugreg6, 6); - set_debugreg(next->debugreg7, 7); - } - if (test_tsk_thread_flag(prev_p, TIF_NOTSC) ^ test_tsk_thread_flag(next_p, TIF_NOTSC)) { /* prev and next are different */ Index: arch/x86/kernel/process_32.c =================================================================== --- arch/x86/kernel/process_32.c.orig 2009-04-01 20:53:43.000000000 +0530 +++ arch/x86/kernel/process_32.c 2009-04-01 20:54:39.000000000 +0530 @@ -61,6 +61,8 @@ #include #include #include +#include +#include asmlinkage void ret_from_fork(void) __asm__("ret_from_fork"); @@ -265,7 +267,14 @@ task_user_gs(p) = get_user_gs(regs); + p->thread.io_bitmap_ptr = NULL; + tsk = current; + err = -ENOMEM; + if (unlikely(test_tsk_thread_flag(tsk, TIF_DEBUG))) + if (copy_thread_hw_breakpoint(tsk, p, clone_flags)) + goto out; + if (unlikely(test_tsk_thread_flag(tsk, TIF_IO_BITMAP))) { p->thread.io_bitmap_ptr = kmemdup(tsk->thread.io_bitmap_ptr, IO_BITMAP_BYTES, GFP_KERNEL); @@ -285,10 +294,13 @@ err = do_set_thread_area(p, -1, (struct user_desc __user *)childregs->si, 0); +out: if (err && p->thread.io_bitmap_ptr) { kfree(p->thread.io_bitmap_ptr); p->thread.io_bitmap_max = 0; } + if (err) + flush_thread_hw_breakpoint(p); ds_copy_thread(p, current); @@ -426,6 +438,25 @@ lazy_load_gs(next->gs); percpu_write(current_task, next_p); + /* + * There's a problem with moving the switch_to_thread_hw_breakpoint() + * call before current is updated. Suppose a kernel breakpoint is + * triggered in between the two. The hw-breakpoint handler will see + * that current is different from the task pointer stored in + * last_debugged_task, so it will think the task pointer is leftover + * from an old task (lazy switching) and will erase it. Then until the + * next context switch, no user-breakpoints will be installed. + * + * The real problem is that it's impossible to update both current and + * last_debugged_task at the same instant, so there will always be a + * window in which they disagree and a breakpoint might get triggered. + * Since we use lazy switching, we are forced to assume that a + * disagreement means that current is correct and last_debugged_task is + * old. But if you move the code above then you'll create a window in + * which current is old and last_debugged_task is correct. + */ + if (unlikely(test_tsk_thread_flag(next_p, TIF_DEBUG))) + switch_to_thread_hw_breakpoint(next_p); return prev_p; } Index: arch/x86/kernel/process_64.c =================================================================== --- arch/x86/kernel/process_64.c.orig 2009-04-01 20:53:43.000000000 +0530 +++ arch/x86/kernel/process_64.c 2009-04-01 20:54:39.000000000 +0530 @@ -55,6 +55,8 @@ #include #include #include +#include +#include asmlinkage extern void ret_from_fork(void); @@ -248,6 +250,8 @@ BUG(); } } + if (unlikely(dead_task->thread.debugreg7)) + flush_thread_hw_breakpoint(dead_task); } static inline void set_32bit_tls(struct task_struct *t, int tls, u32 addr) @@ -303,12 +307,18 @@ p->thread.fs = me->thread.fs; p->thread.gs = me->thread.gs; + p->thread.io_bitmap_ptr = NULL; savesegment(gs, p->thread.gsindex); savesegment(fs, p->thread.fsindex); savesegment(es, p->thread.es); savesegment(ds, p->thread.ds); + err = -ENOMEM; + if (unlikely(test_tsk_thread_flag(me, TIF_DEBUG))) + if (copy_thread_hw_breakpoint(me, p, clone_flags)) + goto out; + if (unlikely(test_tsk_thread_flag(me, TIF_IO_BITMAP))) { p->thread.io_bitmap_ptr = kmalloc(IO_BITMAP_BYTES, GFP_KERNEL); if (!p->thread.io_bitmap_ptr) { @@ -346,6 +356,9 @@ kfree(p->thread.io_bitmap_ptr); p->thread.io_bitmap_max = 0; } + if (err) + flush_thread_hw_breakpoint(p); + return err; } @@ -491,6 +504,26 @@ */ if (tsk_used_math(next_p) && next_p->fpu_counter > 5) math_state_restore(); + /* + * There's a problem with moving the switch_to_thread_hw_breakpoint() + * call before current is updated. Suppose a kernel breakpoint is + * triggered in between the two. The hw-breakpoint handler will see + * that current is different from the task pointer stored in + * last_debugged_task, so it will think the task pointer is leftover + * from an old task (lazy switching) and will erase it. Then until the + * next context switch, no user-breakpoints will be installed. + * + * The real problem is that it's impossible to update both current and + * last_debugged_task at the same instant, so there will always be a + * window in which they disagree and a breakpoint might get triggered. + * Since we use lazy switching, we are forced to assume that a + * disagreement means that current is correct and last_debugged_task is + * old. But if you move the code above then you'll create a window in + * which current is old and last_debugged_task is correct. + */ + if (unlikely(test_tsk_thread_flag(next_p, TIF_DEBUG))) + switch_to_thread_hw_breakpoint(next_p); + return prev_p; } Index: arch/x86/include/asm/processor.h =================================================================== --- arch/x86/include/asm/processor.h.orig 2009-04-01 20:53:46.000000000 +0530 +++ arch/x86/include/asm/processor.h 2009-04-01 20:54:39.000000000 +0530 @@ -426,10 +426,6 @@ unsigned long fs; unsigned long gs; /* Hardware debugging registers: */ - unsigned long debugreg0; - unsigned long debugreg1; - unsigned long debugreg2; - unsigned long debugreg3; unsigned long debugreg[HB_NUM]; unsigned long debugreg6; unsigned long debugreg7; -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/