Re: [PATCH v30 07/20] x86/sgx: Enumerate and track EPC sections

[Jarkko Sakkinen]

On Thu, May 28, 2020 at 09:14:43AM +0300, Jarkko Sakkinen wrote:
> On Thu, May 28, 2020 at 08:35:15AM +0300, Jarkko Sakkinen wrote:
> > On Thu, May 28, 2020 at 08:25:43AM +0300, Jarkko Sakkinen wrote:
> > > On Tue, May 26, 2020 at 08:56:14PM -0700, Sean Christopherson wrote:
> > > > On Mon, May 25, 2020 at 11:23:04AM +0200, Borislav Petkov wrote:
> > > > > On Fri, May 15, 2020 at 03:43:57AM +0300, Jarkko Sakkinen wrote:
> > > > > > +struct sgx_epc_section sgx_epc_sections[SGX_MAX_EPC_SECTIONS];
> > > > > > +int sgx_nr_epc_sections;
> > > > > 
> > > > > We have become very averse against global stuff. What is going to use
> > > > > those, only sgx code I assume...?
> > > > 
> > > > Yes, only SGX code.  The reclaim/swap code needs access to the sections,
> > > > and that code is in a different file, reclaim.c.  I don't have a super
> > > > strong objection to sucking reclaim.c into main.c, but I'm somewhat
> > > > indifferent on code organization as a whole.  Jarkko likely has a stronger
> > > > opinion.
> > > 
> > > I'll change it.
> > > 
> > > It's not quite as easy as just "sucking the file in". All the commits
> > > that touch the file need to be reworked:
> > > 
> > > $ git --no-pager log --format="%H %s" arch/x86/kernel/cpu/sgx/reclaim.c
> > > 5aeca6dabf767e9350ee3188ba25ceb21f3162b4 x86/sgx: Add a page reclaimer
> > > de9b1088959f36ffdaf43a49bfea1c7f9f81cac7 x86/sgx: Linux Enclave Driver
> > > 08d8fcb74fe268059ee58fcc2a0833b244e1f22a x86/sgx: Enumerate and track EPC sections
> > 
> > Not that I haven't done this a lot last few years. A proven approach
> > is to do it in two "git rebase -i mainline/master" sweeps:
> > 
> > 1. For each commit, remove reclaim.c entry from the Makefile and import
> >    reclaim.c contents to main.c.
> > 2. For each commit, delete reclaim.c.
> > 
> > I've tried quite a few different angles and this what I've converged
> > into. Very hard to hit messy into messy merge conflicts.
> 
> Remembered why the things are the way they are. Also ioctl.c needs these
> symbols and I'd keep that separate from the contents of main.c and
> reclaim.c. There the separation obviously makes sense.
> 
> I'll anyway merge main.c and reclaim.c as one for v31 because they are
> strongly connected.

And more importantly for the reason that it allows to make ksgxswapd_tsk
making the whole thing way more cleaner.

/Jarkko