Re: [PATCH 1/2] KVM: selftests: Make rseq compatible with glibc-2.35

From: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: Florian Weimer <fweimer@redhat.com>,
	shan gavin <shan.gavin@gmail.com>, KVM list <kvm@vger.kernel.org>,
	maz <maz@kernel.org>, linux-kernel <linux-kernel@vger.kernel.org>,
	andrew jones <andrew.jones@linux.dev>, yihyu <yihyu@redhat.com>,
	linux-kselftest <linux-kselftest@vger.kernel.org>,
	kvmarm <kvmarm@lists.cs.columbia.edu>
Subject: Re: [PATCH 1/2] KVM: selftests: Make rseq compatible with glibc-2.35
Date: Wed, 10 Aug 2022 08:17:36 -0400 (EDT)	[thread overview]
Message-ID: <392275844.358.1660133856396.JavaMail.zimbra@efficios.com> (raw)
In-Reply-To: <bb97efaf-4f58-c192-a489-e71ebbebce8c@redhat.com>

----- On Aug 10, 2022, at 5:14 AM, Paolo Bonzini pbonzini@redhat.com wrote:

> On 8/9/22 14:21, Mathieu Desnoyers wrote:
>>> For kvm/selftests, there are 3 architectures involved actually. So we
>>> just need consider 4 cases: aarch64, x86, s390 and other. For other
>>> case, we just use __builtin_thread_pointer() to maintain code's
>>> integrity, but it's not called at all.
>>>
>>> I think kvm/selftest is always relying on glibc if I'm correct.
>> All those are handled in the rseq selftests and in librseq. Why duplicate all
>> that logic again?
> 
> Yeah, rseq_test should reuse librseq code.  The simplest way,
> if slightly hackish, is to do something like
> 
> diff --git a/tools/testing/selftests/kvm/Makefile
> b/tools/testing/selftests/kvm/Makefile
> index 690b499c3471..6c192b0ec304 100644
> --- a/tools/testing/selftests/kvm/Makefile
> +++ b/tools/testing/selftests/kvm/Makefile
> @@ -37,6 +37,7 @@ ifeq ($(ARCH),riscv)
>  	UNAME_M := riscv
>  endif
>  
>  LIBKVM += lib/assert.c
>  LIBKVM += lib/elf.c
>  LIBKVM += lib/guest_modes.c
> @@ -198,7 +199,7 @@ endif
>  CFLAGS += -Wall -Wstrict-prototypes -Wuninitialized -O2 -g -std=gnu99 \
>  	-fno-stack-protector -fno-PIE -I$(LINUX_TOOL_INCLUDE) \
>  	-I$(LINUX_TOOL_ARCH_INCLUDE) -I$(LINUX_HDR_PATH) -Iinclude \
> -	-I$(<D) -Iinclude/$(UNAME_M) -I.. $(EXTRA_CFLAGS) $(KHDR_INCLUDES)
> +	-I$(<D) -Iinclude/$(UNAME_M) -I.. $(EXTRA_CFLAGS) $(KHDR_INCLUDES) -I../rseq
>  
>  no-pie-option := $(call try-run, echo 'int main() { return 0; }' | \
>          $(CC) -Werror -no-pie -x c - -o "$$TMP", -no-pie)
> 
> 
> and just #include "../rseq/rseq.c" in rseq_test.c.

Hi Paolo,

Indeed, this hack seems to be a good approach to immediately fix things without
moving around all source files and headers. In the longer term, I'd prefer Sean's
proposal to move rseq.c to tools/lib/ (and to move rseq headers to tools/include/rseq/).
This can be done in a follow up phase though. I'll put a note on my todo list
for after I come back from vacation.

I'll be able to do this refactoring on top of this fix.

Thanks,

Mathieu

> 
> Thanks,
> 
> Paolo

-- 
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm