linux-kernel - Re: [PATCH 07/15] powerpc/watchpoint: Get watchpoint count dynamically while disabling them

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-Id: <a7cde9ea-02d6-a5f9-3124-6047b4e1c622@linux.ibm.com>
Date:   Thu, 26 Mar 2020 09:02:05 +0530
From:   Ravi Bangoria <ravi.bangoria@...ux.ibm.com>
To:     Christophe Leroy <christophe.leroy@....fr>
Cc:     mpe@...erman.id.au, mikey@...ling.org, apopple@...ux.ibm.com,
        paulus@...ba.org, npiggin@...il.com,
        naveen.n.rao@...ux.vnet.ibm.com, peterz@...radead.org,
        jolsa@...nel.org, oleg@...hat.com, fweisbec@...il.com,
        mingo@...nel.org, linuxppc-dev@...ts.ozlabs.org,
        linux-kernel@...r.kernel.org,
        Ravi Bangoria <ravi.bangoria@...ux.ibm.com>
Subject: Re: [PATCH 07/15] powerpc/watchpoint: Get watchpoint count
 dynamically while disabling them



On 3/18/20 12:27 PM, Ravi Bangoria wrote:
> 
> 
> On 3/17/20 4:02 PM, Christophe Leroy wrote:
>>
>>
>> Le 09/03/2020 à 09:57, Ravi Bangoria a écrit :
>>> Instead of disabling only one watchpooint, get num of available
>>> watchpoints dynamically and disable all of them.
>>>
>>> Signed-off-by: Ravi Bangoria <ravi.bangoria@...ux.ibm.com>
>>> ---
>>>   arch/powerpc/include/asm/hw_breakpoint.h | 15 +++++++--------
>>>   1 file changed, 7 insertions(+), 8 deletions(-)
>>>
>>> diff --git a/arch/powerpc/include/asm/hw_breakpoint.h b/arch/powerpc/include/asm/hw_breakpoint.h
>>> index 980ac7d9f267..ec61e2b7195c 100644
>>> --- a/arch/powerpc/include/asm/hw_breakpoint.h
>>> +++ b/arch/powerpc/include/asm/hw_breakpoint.h
>>> @@ -75,14 +75,13 @@ extern void ptrace_triggered(struct perf_event *bp,
>>>               struct perf_sample_data *data, struct pt_regs *regs);
>>>   static inline void hw_breakpoint_disable(void)
>>>   {
>>> -    struct arch_hw_breakpoint brk;
>>> -
>>> -    brk.address = 0;
>>> -    brk.type = 0;
>>> -    brk.len = 0;
>>> -    brk.hw_len = 0;
>>> -    if (ppc_breakpoint_available())
>>> -        __set_breakpoint(&brk, 0);
>>> +    int i;
>>> +    struct arch_hw_breakpoint null_brk = {0};
>>> +
>>> +    if (ppc_breakpoint_available()) {
>>
>> I think this test should go into nr_wp_slots() which should return zero when no breakpoint is available.
> 
> Seems possible. Will change it in next version.

Once we move ppc_breakpoint_available() logic into nr_wp_slots(),
'dawr_force_enable' variable is checked in nr_wp_slots() before
it gets initialized in dawr_force_setup():

     start_kernel()
     |-> perf_event_init()
     |     init_hw_breakpoint()
     |       hw_breakpoint_slots()
     |         nr_wp_slots()
     |           /* Check dawr_force_enable variable */
     |
     |-> arch_call_rest_init()
           rest_init()
             kernel_thread(kernel_init, ...)
               ...
                 kernel_init()
                   kernel_init_freeable()
                     ...
                     do_one_initcall()
                       dawr_force_setup()
                         /* Set dawr_force_enable = true */

Because of this, hw-breakpoint infrastructure is initialized with
no DAWRs. So I'm thinking to keep the code as it is i.e. not moving
ppc_breakpoint_available() test inside nr_wp_slots().

Ravi