Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Jiri Olsa <olsajiri@gmail.com>
Date:   Tue, 28 Jun 2022 17:35:21 +0200
To:     Ian Rogers <irogers@google.com>
Cc:     olsajiri@gmail.com, Peter Zijlstra <peterz@infradead.org>,
        Ingo Molnar <mingo@redhat.com>,
        Arnaldo Carvalho de Melo <acme@kernel.org>,
        Mark Rutland <mark.rutland@arm.com>,
        Alexander Shishkin <alexander.shishkin@linux.intel.com>,
        Namhyung Kim <namhyung@kernel.org>,
        Alexei Starovoitov <ast@kernel.org>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Andrii Nakryiko <andrii@kernel.org>,
        Martin KaFai Lau <kafai@fb.com>,
        Song Liu <songliubraving@fb.com>, Yonghong Song <yhs@fb.com>,
        John Fastabend <john.fastabend@gmail.com>,
        KP Singh <kpsingh@kernel.org>,
        Dave Marchevsky <davemarchevsky@fb.com>,
        Quentin Monnet <quentin@isovalent.com>,
        linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org,
        netdev@vger.kernel.org, bpf@vger.kernel.org,
        Stephane Eranian <eranian@google.com>
Subject: Re: [PATCH] perf bpf: 8 byte align bpil data
Message-ID: <YrsfueiaxKmpf0Ng@krava>
References: <20220614014714.1407239-1-irogers@google.com>
 <Yrq5Bun3Nmb1vrW3@krava>
 <CAP-5=fXNJjRxGCE=mH22bLg1mNXMRgL_px4=-=8Zq-DLUXbxTg@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAP-5=fXNJjRxGCE=mH22bLg1mNXMRgL_px4=-=8Zq-DLUXbxTg@mail.gmail.com>
Precedence: bulk

On Tue, Jun 28, 2022 at 08:15:04AM -0700, Ian Rogers wrote:
> On Tue, Jun 28, 2022 at 1:41 AM <olsajiri@gmail.com> wrote:
> >
> > On Mon, Jun 13, 2022 at 06:47:14PM -0700, Ian Rogers wrote:
> > > bpil data is accessed assuming 64-bit alignment resulting in undefined
> > > behavior as the data is just byte aligned. With an -fsanitize=undefined
> > > build the following errors are observed:
> >
> > I need to add -w to get the clean build with that, do you see that as well?
> >
> >   $ make EXTRA_CFLAGS='-fsanitize=undefined -w'
> 
> I don't recall needing this, but I was stacking fixes which may explain it.
> 
> > >
> > > $ sudo perf record -a sleep 1
> > > util/bpf-event.c:310:22: runtime error: load of misaligned address 0x55f61084520f for type '__u64', which requires 8 byte alignment
> > > 0x55f61084520f: note: pointer points here
> > >  a8 fe ff ff 3c  51 d3 c0 ff ff ff ff 04  84 d3 c0 ff ff ff ff d8  aa d3 c0 ff ff ff ff a4  c0 d3 c0
> > >              ^
> > > util/bpf-event.c:311:20: runtime error: load of misaligned address 0x55f61084522f for type '__u32', which requires 4 byte alignment
> > > 0x55f61084522f: note: pointer points here
> > >  ff ff ff ff c7  17 00 00 f1 02 00 00 1f  04 00 00 58 04 00 00 00  00 00 00 0f 00 00 00 63  02 00 00
> > >              ^
> > > util/bpf-event.c:198:33: runtime error: member access within misaligned address 0x55f61084523f for type 'const struct bpf_func_info', which requires 4 byte alignment
> > > 0x55f61084523f: note: pointer points here
> > >  58 04 00 00 00  00 00 00 0f 00 00 00 63  02 00 00 3b 00 00 00 ab  02 00 00 44 00 00 00 14  03 00 00
> >
> >
> > and I'm also getting another error in:
> >
> > [root@krava perf]# ./perf record -a sleep 1
> > util/synthetic-events.c:1202:11: runtime error: member access within misaligned address 0x00000286f7ea for type 'struct perf_record_record_cpu_map', which requires 8 byte alignment
> > 0x00000286f7ea: note: pointer points here
> >  20 00  01 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  00 00 00 00 00 00
> >               ^
> > util/synthetic-events.c:1203:18: runtime error: member access within misaligned address 0x00000286f7ea for type 'struct perf_record_record_cpu_map', which requires 8 byte alignment
> > 0x00000286f7ea: note: pointer points here
> >  20 00  01 00 01 00 00 00 00 00  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  00 00 00 00 00 00
> >               ^
> > util/synthetic-events.c:1206:46: runtime error: member access within misaligned address 0x00000286f7ea for type 'struct perf_record_record_cpu_map', which requires 8 byte alignment
> > 0x00000286f7ea: note: pointer points here
> >  20 00  01 00 01 00 08 00 00 00  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  00 00 00 00 00 00
> >               ^
> > /home/jolsa/kernel/linux-perf/tools/include/asm-generic/bitops/atomic.h:10:29: runtime error: load of misaligned address 0x00000286f7f2 for type 'long unsigned int', which requires 8 byte alignment
> > 0x00000286f7f2: note: pointer points here
> >  00 00  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  51 00 00 00 00 00
> >               ^
> >
> > are you going to address this one as well?
> >
> >
> > the reason for this one is that 'data' in struct perf_record_cpu_map_data
> > is not alligned(8), so that's why I raised the question in my other reply ;-)
> >
> > I wonder we should mark all tools/lib/perf/include/perf/event.h types
> > as packed to prevent any compiler padding
> 
> I already sent out a fix and some improvements related to this:
> https://lore.kernel.org/lkml/20220614143353.1559597-1-irogers@google.com/
> Could you take a look?

ok, I overlooked that one

> 
> I'm not sure about aligned and packed. I tried to minimize it in the
> change above. The issue is that taking the address of a variable in a
> packed struct results in an unaligned pointer. To address this in the
> fix above I changed the functions to pass pointers to the whole
> struct.

ok, will check,

thanks,
jirka