Re: [PATCH 1/3] LoongArch: tools: Add relocs tool support

[Youling Tang]

Hi, Jinyang

Thank you very much for some ideas.

On 09/04/2022 12:23 AM, Jinyang He wrote:
> On 2022/9/3 18:49, Xi Ruoyao wrote:
>
> > On Sat, 2022-09-03 at 09:57 +0800, Youling Tang wrote:
> > > > > Unlike (pre-r6) MIPS, LoongArch has a complete support for PIC, and
> > > > > currently LoongArch toolchain always produces PIC (except, if
> > > > > -Wa,-mla-
> > > > > {local,global}-with-abs or la.abs macros are used explicitly).
> > > > >
> > > > > So would it be easier to review and correct the uses of "la.abs" in
> > > > > the
> > > > > code, and make the main kernel image a real PIE?  Then we can load it
> > > > > everywhere w/o any need to do relocation at load time.
> > > At the beginning I also wanted to make the main kernel image a real PIE
> > > and tried it, some of the "la.abs" can be modified, but I encountered
> > > difficulties in modifying the  exception handling code part, the kernel
> > > will not  boot after modification :(, I will continue to work hard try.
> > I just tried the same thing and get the same result :(.  Will spend
> > several hours reading the LoongArch manual about exception...
> Hi,
>
> The following ideas are based on experience, without validation. Patches
> show that three types of relocation are needed to be done.
> 1, GOT is generated by toolchain, so  I think eliminating them by
> toolchain is better.
> 2, Ex_table is generated but striped relocation info. We can plays pcrel
> way to resolve this problem. One of ways like follows, (pseudo-code)
>
> arch/loongarch/copy_user.S
> ...
> .macro fixup_ex from, to, offset, fix
> .if \fix
>         .section .fixup, "ax"
> \to:    addi.d  a0, a2, \offset
>         jr      ra
>         .previous
> .endif
>         .section __ex_table, "a"
>         PTR     \from\()b, \to\()b - \from\()b   # fixup use pcrel
>         .previous
> .endm
> ...
> ...
>
> arch/loongarch/mm/extable.c
>         ...
>         unsigned long era = exception_era(regs);
>         fixup = search_exception_tables(era - relocate_offset); //
> Search by orig value
>         if (fixup) {
>                 regs->csr_era = fixup->fixup + era; // Fixup fixup address.
>                 return 1;
>         }
>
>         ...
Yes, we might be able to make the exception table consists of pairs of
relative offsets, like arm64 and riscv do.

Thanks,
Youling.
> 3, The left is the processed relocation table. It collect the relocation
> type R_LARCH_ABS* (R_LARCH_MARK_LA?) and R_LARCH_64. Just do
> as this patch does. The la.abs is used because of the different exception
> entry on each CPUs and better NUMA performance IMHO. We can
> do follows way to avoid R_LARCH_ABS* and produce R_LARCH_64, (data in text)
>
> .text
> la.pcrel t0,999f
> ld.d t0,t0,0
> ...
> PTR do_vint
> ...
>
> *Not* important. For me, I really wish a bootstrap relocation while this
> patch is KALSR. The bootstrap relocation meaning is the boot cpu enter
> kernel_entry, check the real address itself, parse the cmdline whether
> do kaslr, and then relocate itself. Then kdump capture process is easier.
> Some past info links,
> https://lore.kernel.org/loongarch/3002b2b3-2c2c-e32e-c474-d3876d3ef9b7@xxxxxxxxxxx/
>
> https://patchwork.kernel.org/project/linux-mips/patch/1618230494-6207-6-git-send-email-hejinyang@xxxxxxxxxxx/