Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH] tracing/x86: fix trace event registration for syscalls
 without arguments
To:     Thomas Gleixner <tglx@linutronix.de>,
        Brian Gerst <brgerst@gmail.com>, linux-kernel@vger.kernel.org
Cc:     Dominik Brodowski <linux@dominikbrodowski.net>,
        Andy Lutomirski <luto@kernel.org>
References: <158636958997.7900.16485049455470033557.stgit@buzz>
 <87368pq97n.fsf@nanos.tec.linutronix.de>
From:   Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Message-ID: <1727d83e-5a3d-952e-d7e8-0e3d52bbafa2@yandex-team.ru>
Date:   Mon, 27 Apr 2020 17:55:50 +0300
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.7.0
MIME-Version: 1.0
In-Reply-To: <87368pq97n.fsf@nanos.tec.linutronix.de>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: en-CA
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


On 27/04/2020 17.44, Thomas Gleixner wrote:
> Konstantin Khlebnikov <khlebnikov@yandex-team.ru> writes:
> 
>> Syscalls without arguments now has no ABI subs, instead of that macro
> 
> What is 'ABI subs'?

typo. ABI stubs

That commmit d2b5de495ee9 ("x86/entry: Refactor SYSCALL_DEFINE0 macros")

replaced usual wrapper

long __x64_syscall()
{	
	return __do_sys_syscall();
}

with alias

long __x64_syscall() __alias__(__do_sys_syscall);

And now symbol lookup for entry point return "__do_sys_syscall"
rather than "__x64_syscall" and ftrace magic cannot determine
syscall number for entry point and vise versa.

> 
>> SYSCALL_DEFINE0() defines __abi_sys_name as aliase to __do_sys_name.
>>
>> As a result in find_syscall_meta() kallsyms_lookup() returns
>> "__do_sys_name" which does not match with declared trace event.
>>
>> Also see commit 1c758a2202a6 ("tracing/x86: Update syscall trace events
>> to handle new prefixed syscall func names")
>>
>> Fixes: d2b5de495ee9 ("x86/entry: Refactor SYSCALL_DEFINE0 macros")
>> Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
>> ---
>>   arch/x86/include/asm/ftrace.h |    5 +++--
>>   1 file changed, 3 insertions(+), 2 deletions(-)
>>
>> diff --git a/arch/x86/include/asm/ftrace.h b/arch/x86/include/asm/ftrace.h
>> index 85be2f506272..70b96cae5b42 100644
>> --- a/arch/x86/include/asm/ftrace.h
>> +++ b/arch/x86/include/asm/ftrace.h
>> @@ -61,11 +61,12 @@ static inline bool arch_syscall_match_sym_name(const char *sym, const char *name
>>   {
>>   	/*
>>   	 * Compare the symbol name with the system call name. Skip the
>> -	 * "__x64_sys", "__ia32_sys" or simple "sys" prefix.
>> +	 * "__x64_sys", "__ia32_sys", "__do_sys" or simple "sys" prefix.
>>   	 */
>>   	return !strcmp(sym + 3, name + 3) ||
>>   		(!strncmp(sym, "__x64_", 6) && !strcmp(sym + 9, name + 3)) ||
>> -		(!strncmp(sym, "__ia32_", 7) && !strcmp(sym + 10, name + 3));
>> +		(!strncmp(sym, "__ia32_", 7) && !strcmp(sym + 10, name + 3)) ||
>> +		(!strncmp(sym, "__do_sys", 8) && !strcmp(sym + 8, name + 3));
>>   }
>>   
>>   #ifndef COMPILE_OFFSETS