Re: [PATCH 3/5] ebpf: add formal error reporting to all APIs

["Philippe Mathieu-Daudé" <philmd@linaro.org>]

On 6/8/24 17:20, Daniel P. Berrangé wrote:
> On Tue, Aug 06, 2024 at 05:11:55PM +0200, Philippe Mathieu-Daudé wrote:
>> On 6/8/24 16:56, Daniel P. Berrangé wrote:
>>> The eBPF code is currently reporting error messages through trace
>>> events. Trace events are fine for debugging, but they are not to be
>>> considered the primary error reporting mechanism, as their output
>>> is inaccessible to callers.
>>>
>>> This adds an "Error **errp" parameter to all methods which have
>>> important error scenarios to report to the caller.
>>>
>>> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
>>> ---
>>>    ebpf/ebpf_rss.c     | 59 ++++++++++++++++++++++++++++++++++++---------
>>>    ebpf/ebpf_rss.h     | 10 +++++---
>>>    hw/net/virtio-net.c |  7 +++---
>>>    3 files changed, 59 insertions(+), 17 deletions(-)
>>>
>>> diff --git a/ebpf/ebpf_rss.c b/ebpf/ebpf_rss.c
>>> index aa7170d997..59854c8b51 100644
>>> --- a/ebpf/ebpf_rss.c
>>> +++ b/ebpf/ebpf_rss.c
>>> @@ -47,13 +47,14 @@ bool ebpf_rss_is_loaded(struct EBPFRSSContext *ctx)
>>>        return ctx != NULL && (ctx->obj != NULL || ctx->program_fd != -1);
>>>    }
>>> -static bool ebpf_rss_mmap(struct EBPFRSSContext *ctx)
>>> +static bool ebpf_rss_mmap(struct EBPFRSSContext *ctx, Error **errp)
>>>    {
>>>        ctx->mmap_configuration = mmap(NULL, qemu_real_host_page_size(),
>>>                                       PROT_READ | PROT_WRITE, MAP_SHARED,
>>>                                       ctx->map_configuration, 0);
>>>        if (ctx->mmap_configuration == MAP_FAILED) {
>>>            trace_ebpf_error("eBPF RSS", "can not mmap eBPF configuration array");
>>> +        error_setg(errp, "Unable to map eBPF configuration array");
>>>            return false;
>>>        }
>>>        ctx->mmap_toeplitz_key = mmap(NULL, qemu_real_host_page_size(),
>>> @@ -61,6 +62,7 @@ static bool ebpf_rss_mmap(struct EBPFRSSContext *ctx)
>>>                                       ctx->map_toeplitz_key, 0);
>>>        if (ctx->mmap_toeplitz_key == MAP_FAILED) {
>>>            trace_ebpf_error("eBPF RSS", "can not mmap eBPF toeplitz key");
>>> +        error_setg(errp, "Unable to map eBPF toeplitz array");
>>>            goto toeplitz_fail;
>>>        }
>>>        ctx->mmap_indirections_table = mmap(NULL, qemu_real_host_page_size(),
>>> @@ -68,6 +70,7 @@ static bool ebpf_rss_mmap(struct EBPFRSSContext *ctx)
>>>                                       ctx->map_indirections_table, 0);
>>>        if (ctx->mmap_indirections_table == MAP_FAILED) {
>>>            trace_ebpf_error("eBPF RSS", "can not mmap eBPF indirection table");
>>> +        error_setg(errp, "Unable to map eBPF indirection array");
>>
>> Aren't these trace_ebpf_error() calls redundant now?
> 
> Yes & no. Errors propagated up the call stack don't get included in
> any trace output, and so if a caller doesn't log them anywhere they
> can thus be invisible to someone just looking at trace output.
> 
> I could remove them all from the eBPF code though, and put a single
> trace event in the hw/net/virtio-net.c file instead ? Bit of a bike
> shed colouring exercise to decide which is best though.

No problem, I'm fine with this patch.

Note from experience (although pre-existing in this patch), trace
events can be very verbose, and a what makes them powerful is we
can filter particular ones. The following pattern isn't practical
to filter:

   trace_foo_error(const char *error_msg);

While a bit tedious to add, having a single trace event per error
is way more useful.

Regards,

Phil.