linux-kernel - Re: [PATCHv5 4/4] asm-generic/io: Add logging support for MMIO accessors

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

Message-ID: <0cd0bc8c-e3db-b3fb-5be4-c619d1d5d633@quicinc.com>
Date:   Mon, 6 Dec 2021 15:22:51 +0530
From:   Sai Prakash Ranjan <quic_saipraka@...cinc.com>
To:     Arnd Bergmann <arnd@...db.de>
CC:     Will Deacon <will@...nel.org>,
        Catalin Marinas <catalin.marinas@....com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Marc Zyngier <maz@...nel.org>,
        gregkh <gregkh@...uxfoundation.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Linux ARM <linux-arm-kernel@...ts.infradead.org>,
        linux-arm-msm <linux-arm-msm@...r.kernel.org>,
        <quic_psodagud@...cinc.com>
Subject: Re: [PATCHv5 4/4] asm-generic/io: Add logging support for MMIO
 accessors

On 12/6/2021 2:39 PM, Arnd Bergmann wrote:
> On Mon, Dec 6, 2021 at 9:28 AM Sai Prakash Ranjan
> <quic_saipraka@...cinc.com> wrote:
>> +#if IS_ENABLED(CONFIG_TRACE_MMIO_ACCESS) && !(defined(__DISABLE_TRACE_MMIO__))
>> +#include <linux/tracepoint-defs.h>
>> +
>> +DECLARE_TRACEPOINT(rwmmio_write);
>> +DECLARE_TRACEPOINT(rwmmio_read);
>> +
>> +#define rwmmio_tracepoint_active(t) tracepoint_enabled(t)
>> +void log_write_mmio(u64 val, u8 width, volatile void __iomem *addr);
>> +void log_read_mmio(u8 width, const volatile void __iomem *addr);
>> +
>> +#else
>> +
>> +#define rwmmio_tracepoint_active(t) false
>> +static inline void log_write_mmio(u64 val, u8 width, volatile void __iomem *addr) {}
>> +static inline void log_read_mmio(u8 width, const volatile void __iomem *addr) {}
>> +
>> +#endif /* CONFIG_TRACE_MMIO_ACCESS */
>>
>>   /*
>>    * __raw_{read,write}{b,w,l,q}() access memory in native endianness.
>> @@ -149,6 +166,8 @@ static inline u8 readb(const volatile void __iomem *addr)
>>   {
>>          u8 val;
>>
>> +       if (rwmmio_tracepoint_active(rwmmio_read))
>> +               log_read_mmio(8, addr);
>>          __io_br();
>>          val = __raw_readb(addr);
>>          __io_ar(val);
> For readability, it may be nicer to fold the two lines you add for each
> helper into one, such as
>
> void __log_write_mmio(u64 val, u8 width, volatile void __iomem *addr);
> #define log_write_mmio(val, widtg, addr) do { \
>       if (tracepoint_enabled(rwmmio_read)) \
>                 __log_write_mmio((val), (width), (addr)); \
> } while (0)
>
> I wonder if it may even be better to not check for tracepoint_active() in the
> inline function at all but always enter the external function when built-in.
> This means we do run into the branch, but it also reduces the i-cache footprint.

Right, we don't need the tracepoint active check as we declare the 
tracepoint only when the config is
enabled so I can just call log_{read,write}_mmio directly.

> For general functionality, I think it would be better to trace the returned
> value from the read, but I don't know if that defeats the purpose you
> are interested in, since it requires the tracing to come after the __raw_read.
>
>         

Yes just the trace after read/write won't serve our usecase where we 
expect crashes/hangs on accessing
these registers but internally we did have a log_post_read_mmio() as 
well, if it is useful then I can add it.

Thanks,
Sai