[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <00c51e7c-165f-17eb-aba5-8bcb780325d7@cs.kuleuven.be>
Date: Mon, 21 Aug 2023 15:24:16 +0200
From: Jo Van Bulck <jo.vanbulck@...kuleuven.be>
To: "Huang, Kai" <kai.huang@...el.com>,
"linux-sgx@...r.kernel.org" <linux-sgx@...r.kernel.org>,
"jarkko@...nel.org" <jarkko@...nel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Cc: "dave.hansen@...ux.intel.com" <dave.hansen@...ux.intel.com>
Subject: Re: [PATCH 2/5] selftests/sgx: Fix function pointer relocation in
test enclave.
On 21.08.23 13:04, Huang, Kai wrote:
> Thanks. You mentioned this was generated by "-static -fPIC -Os" but I think
> using "-static-pie -fPIE -Os" would probably generate the same.
Thanks, yes, I can confirm that the same code is generated with
-static-pie -FPIE.
> Right. Even for "-static-pie" it's still possible to generate "rela.dyn" for
> those have to use absolute addressing.
Indeed.
>> Thus, my earlier comment says that we can safely ignore these
>> apparent/false TCS "relocations".
>
> Yeah. I guess that's why the test_encl.elf must be built starting from address
> 0.
True (for the ELF in the linker script, but we still need the runtime
relocations for the function-pointer table when loading the enclave at a
non-zero base address).
> Yeah I guess we should just use "-fPIE".
Agreed.
Best,
Jo
Powered by blists - more mailing lists