<div dir="auto"><div><br><br><div class="gmail_quote"><div dir="ltr">On Fri, May 25, 2018, 12:11 PM Amaan Cheval <<a href="mailto:amaan.cheval@gmail.com">amaan.cheval@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hey! Could you link me to some code that you used for the Deos setup<br>
you mentioned?<br>
My understanding is that the -shared option can link static archives<br>
to create a "shared" library in the sense that it doesn't include the<br>
usual crt0 runtime environment and whatnot, but the code within is<br>
still position-dependent. Given that the PE image that EFI needs is<br>
one that needs to be truly relocatable, this may not work - BUT, I've<br>
only just noticed the ./gnuefi/reloc_x86_64.c file which seems to<br>
handle some kinds of runtime relocations encoded within the converted<br>
PE file, so maybe this will work after all. I'll continue to<br>
investigate and let you know how it goes!<br></blockquote></div></div><div dir="auto"><br></div><div dir="auto">Deos isn't a good example except that you can compile with -fPIC and put that code into a static library. Deos is a closed source Level A (man rated flight) ARINC 653 RTOS. It's boot process reads configuration information about each partition and associates .so's with each address space per the configuration. It can't change after that.</div><div dir="auto"><br></div><div dir="auto">The RTEMS exe is mostly linked as normal except to use some arguments to say some symbols are from a shared library.</div><div dir="auto"><br></div><div dir="auto">The base address of the exe is that of the provided virtual address space with .data and .bss in their respective spaces.</div><div dir="auto"><br></div><div dir="auto">And our entry point is in C so there is no asm before that. Great simplification.</div><div dir="auto"><br></div><div dir="auto"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
Regarding how TLS differs with PIC - could you elaborate? Is it<br>
something we'll need to solve for if we go with the -fPIC option, or<br>
is it something we need to keep in mind as a limitation, but isn't<br>
really a blocker?<br></blockquote></div></div><div dir="auto"><br></div><div dir="auto">I don't think PIC changes the TLS mechanism on arm or PowerPC but on i386, when not PIC the TLS base is in %gs and it's a subroutine call when PIC. Just as well for Deos since they assume an application won't change the segment register values.</div><div dir="auto"><br></div><div dir="auto">Other than this one TLS difference, it is a normal exe to me. They just magically provide their .so's before we run.</div><div dir="auto"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
On Fri, May 25, 2018 at 10:13 PM, Joel Sherrill <<a href="mailto:joel@rtems.org" target="_blank" rel="noreferrer">joel@rtems.org</a>> wrote:<br>
><br>
><br>
> On Fri, May 25, 2018, 11:15 AM Amaan Cheval <<a href="mailto:amaan.cheval@gmail.com" target="_blank" rel="noreferrer">amaan.cheval@gmail.com</a>> wrote:<br>
>><br>
>> Hey!<br>
>><br>
>> Skippable details about how FreeBSD handles the UEFI boot process!<br>
>><br>
>> --------------------------------------------------------------------------------<br>
>><br>
>> Having looked into it a bit more, my understanding of how FreeBSD<br>
>> handles this process is:<br>
>> - They build a two-stage bootloader for EFI, called boot1.efi and<br>
>> loader.efi[1]<br>
>> - loader.efi is an interactive prompt which may autoboot, or a "boot<br>
>> kernelImg" command can be used to load the actual kernel<br>
>> - The kernel is loaded as an ELF through helper functions. The<br>
>> command_boot[2] function drives this:<br>
>> - In brief, through calls go through:<br>
>> command_boot -> mod_loadkld -> file_load -><br>
>> file_formats[i]->l_load (actually the loadfile function in<br>
>> load_elf.c[3])<br>
>> - The loadfile function parses the program and section headers of<br>
>> the ELF file (through more function detours that are not really<br>
>> important)<br>
>> - Once the ELF has been loaded at the correct entry_addr that it<br>
>> expects to be loaded at in memory, the l_exec[4] function is called,<br>
>> which is actually elf64_exec in elf64_freebsd.c[5], at which hopefully<br>
>> through trampolining magic, the control flow will transfer to the<br>
>> kernel or ELF module<br>
>><br>
>><br>
>> --------------------------------------------------------------------------------<br>
>><br>
>> What this means for RTEMS if we go with gnu-efi is essentially 2<br>
>> options, given that the objcopy method of converting from ELF->PE<br>
>> requires the ELF to be a position-independent shared library:<br>
>><br>
>> - Using -fPIC to compile all of RTEMS, including the RTEMS user's<br>
>> application code. This way we'd have librtemsbso.so, librtemscpu.so,<br>
>> etc. which would then be linked into user_app.c through -fPIC and<br>
>> -shared flags still, creating one singular hello.so, which can then<br>
>> finally be converted into hello.efi and put on a FAT filesystem and<br>
>> booted. This seems doable, but I'm fairly concerned about it further<br>
>> complicating our build system and likely being quite singular in its<br>
>> focus on EFI.<br>
><br>
><br>
> I'm using PIC on the Deos BSP. RTEMS is still a .a and exes are linked with<br>
> our static libraries and Deos .so.<br>
><br>
> Unless the loader forces something, you can use PIC with no build system<br>
> changes.<br>
><br>
> note that thread local storage is different on i386 with and without PIC.<br>
>><br>
>><br>
>> - The FreeBSD way of a (loader.efi) and a hello.exe (ELF64) put on<br>
>> possibly the same partition on the FAT filesystem required for UEFI<br>
>> application images anyway. The loader.efi can find the hello.exe file<br>
>> through perhaps a config file it can read or by having a magic-name<br>
>> like rtems.exe or something. This effectively means we need an ELF<br>
>> dynamic linker / loader (akin to ld.so) within RTEMS' source. I think<br>
>> using FreeBSD's code for this should be fine. One added benefit of<br>
>> this method is that librtems* and user applications remain as ELF64s,<br>
>> which in the future could also be used with Multiboot with a slightly<br>
>> modified "loader" (i.e. one which generates the apt Multiboot magic<br>
>> header, and boots the PC from 32-bit protected mode to 64-bit long<br>
>> mode).<br>
>><br>
>> I prefer the latter approach personally. If both of these seem too<br>
>> complicated, we can of course go back to considering generating the PE<br>
>> header format in ASM the way Linux distros use EFISTUB and the code<br>
>> Chris shared (as I mentioned in my original blog post) for wimboot.<br>
>> Those approaches may be significantly simpler in a sense, but may<br>
>> limit how we use UEFI Services - I'm not sure about the details of<br>
>> this yet - I can investigate if y'all aren't fond of the option I laid<br>
>> down above.<br>
>><br>
>> Let me know!<br>
>><br>
>> [1]<br>
>> <a href="https://www.freebsd.org/cgi/man.cgi?query=loader&apropos=0&sektion=8&manpath=FreeBSD+11.1-RELEASE+and+Ports&arch=default&format=html" rel="noreferrer noreferrer" target="_blank">https://www.freebsd.org/cgi/man.cgi?query=loader&apropos=0&sektion=8&manpath=FreeBSD+11.1-RELEASE+and+Ports&arch=default&format=html</a><br>
>> [2]<br>
>> <a href="https://github.com/freebsd/freebsd/blob/433bd38e3a0349f9f89f9d54594172c75b002b74/stand/common/boot.c#L53" rel="noreferrer noreferrer" target="_blank">https://github.com/freebsd/freebsd/blob/433bd38e3a0349f9f89f9d54594172c75b002b74/stand/common/boot.c#L53</a><br>
>> [3]<br>
>> <a href="https://github.com/freebsd/freebsd/blob/d8596f6f687a64b994b065f3058155405dfc39db/stand/common/load_elf.c#L150" rel="noreferrer noreferrer" target="_blank">https://github.com/freebsd/freebsd/blob/d8596f6f687a64b994b065f3058155405dfc39db/stand/common/load_elf.c#L150</a><br>
>> [4]<br>
>> <a href="https://github.com/freebsd/freebsd/blob/433bd38e3a0349f9f89f9d54594172c75b002b74/stand/common/boot.c#L107" rel="noreferrer noreferrer" target="_blank">https://github.com/freebsd/freebsd/blob/433bd38e3a0349f9f89f9d54594172c75b002b74/stand/common/boot.c#L107</a><br>
>> [5]<br>
>> <a href="https://github.com/freebsd/freebsd/blob/d8596f6f687a64b994b065f3058155405dfc39db/stand/efi/loader/arch/amd64/elf64_freebsd.c#L93" rel="noreferrer noreferrer" target="_blank">https://github.com/freebsd/freebsd/blob/d8596f6f687a64b994b065f3058155405dfc39db/stand/efi/loader/arch/amd64/elf64_freebsd.c#L93</a><br>
>><br>
>> On Sun, May 20, 2018 at 10:52 PM, Joel Sherrill <<a href="mailto:joel@rtems.org" target="_blank" rel="noreferrer">joel@rtems.org</a>> wrote:<br>
>> ><br>
>> ><br>
>> > On Sun, May 20, 2018, 12:10 PM Amaan Cheval <<a href="mailto:amaan.cheval@gmail.com" target="_blank" rel="noreferrer">amaan.cheval@gmail.com</a>><br>
>> > wrote:<br>
>> >><br>
>> >> On Sat, May 19, 2018 at 6:51 PM, Gedare Bloom <<a href="mailto:gedare@rtems.org" target="_blank" rel="noreferrer">gedare@rtems.org</a>> wrote:<br>
>> >> > On Fri, May 18, 2018 at 5:53 PM, Joel Sherrill <<a href="mailto:joel@rtems.org" target="_blank" rel="noreferrer">joel@rtems.org</a>><br>
>> >> > wrote:<br>
>> >> >><br>
>> >> >><br>
>> >> >> On Fri, May 18, 2018 at 3:24 PM, Amaan Cheval<br>
>> >> >> <<a href="mailto:amaan.cheval@gmail.com" target="_blank" rel="noreferrer">amaan.cheval@gmail.com</a>><br>
>> >> >> wrote:<br>
>> >> >>><br>
>> >> >>> Hi everyone!<br>
>> >> >>><br>
>> >> >>> I've written a quick blog post summarizing the options I've<br>
>> >> >>> considered<br>
>> >> >>> to make the x86_64 port work with UEFI firmware - the primary<br>
>> >> >>> winner<br>
>> >> >>> seems to be in my eyes to use "gnu-efi" and to add support for the<br>
>> >> >>> target "pei-x86-64" (aliased to "efi-app-x86_64") to<br>
>> >> >>> "x86_64-rtems5-objcopy" in binutils. I've submitted a patch for<br>
>> >> >>> this<br>
>> >> >>> here[1].<br>
>> >> >><br>
>> >> >><br>
>> >> >> That patch is quite simple so shouldn't be a problem if this is the<br>
>> >> >> direction<br>
>> >> >> that gets consensus.<br>
>> >> >>><br>
>> >> >>><br>
>> >> >>> The blog post is here:<br>
>> >> >>> <a href="https://blog.whatthedude.com/post/uefi-app-options/" rel="noreferrer noreferrer" target="_blank">https://blog.whatthedude.com/post/uefi-app-options/</a><br>
>> >> >>><br>
>> >> >>> I'd appreciate all feedback (and please do let me know if I haven't<br>
>> >> >>> provided enough context)!<br>
>> >> >>><br>
>> >> >>> Specifically, some concerns I'd like to discuss are:<br>
>> >> >>><br>
>> >> >>> - Does everyone agree with me on choosing gnu-efi + objcopy as our<br>
>> >> >>> method of choice?<br>
>> >> >><br>
>> >> >><br>
>> >> >> Does using gnu-efi add code that runs on the target? Can you point<br>
>> >> >> us to the files, if so.<br>
>> >><br>
>> >> Sure. The files would run on the target, yes. These are the ones<br>
>> >> listed here (as linked to in my blog post, perhaps without sufficient<br>
>> >> emphasis):<br>
>> >> <a href="https://wiki.osdev.org/UEFI#Developing_with_GNU-EFI" rel="noreferrer noreferrer" target="_blank">https://wiki.osdev.org/UEFI#Developing_with_GNU-EFI</a><br>
>> >><br>
>> >> >><br>
>> >> >> Can you tell which approach FreeBSD takes?<br>
>> >><br>
>> >> FreeBSD takes the gnu-efi approach I see as the "winner" here (also a<br>
>> >> link in the post):<br>
>> >><br>
>> >><br>
>> >> <a href="https://github.com/freebsd/freebsd/blob/996b0b6d81cf31cd8d58af5d8b45f0b4945d960d/stand/efi/loader/Makefile#L98-L1" rel="noreferrer noreferrer" target="_blank">https://github.com/freebsd/freebsd/blob/996b0b6d81cf31cd8d58af5d8b45f0b4945d960d/stand/efi/loader/Makefile#L98-L1</a><br>
>> ><br>
>> ><br>
>> > This is (no surprise) appropriately licensed and IMO the winning<br>
>> > solution.<br>
>> > Knowing it is what FreeBSD does makes it an easy choice.<br>
>> ><br>
>> > A comment in the readme mentions there is a i386 version of this code so<br>
>> > that could be used to let pc386 boot from UEFI.<br>
>> ><br>
>> >><br>
>> >><br>
>> >> >><br>
>> >> >>><br>
>> >> >>> - How do we integrate gnu-efi into our build process? A part of the<br>
>> >> >>> RSB, making sure the path to the libraries are in an exported<br>
>> >> >>> variable? Or perhaps a part of the RTEMS kernel itself if the<br>
>> >> >>> licenses<br>
>> >> >>> are compatible (I don't see any on the project[2], only copyright<br>
>> >> >>> notices within the source files of the release versions).<br>
>> >> >><br>
>> >> >><br>
>> >> >> GNU-efi would be built like qemu or the device tree compiler would<br>
>> >> >> be my guess and x86_64-rtems toolset might add that to the standard<br>
>> >> >> set of tools. License on host tools being GPL isn't an issue.<br>
>> >> >><br>
>> >> ><br>
>> >> > It appears to be a standard 2-clause BSD released by Intel as<br>
>> >> > specified in the README file of gnu-efi.<br>
>> >> ><br>
>> >> >><br>
>> >> >>><br>
>> >> >>> - Regardless of how we manage UEFI, do we require Multiboot support<br>
>> >> >>> too? Multiboot drops us in a 32-bit protected mode environment,<br>
>> >> >>> whereas 64-bit UEFI firmware will boot us into 64-bit long mode -<br>
>> >> >>> this<br>
>> >> >>> would mean the kernel would need to support separate code-paths for<br>
>> >> >>> the 2 if we want to support both methods.<br>
>> >> >><br>
>> >> >><br>
>> >> >> That's a good question. For GSoC, I think UEFI is fine and perhaps a<br>
>> >> >> ticket<br>
>> >> >> under the general "modern PC support" ticket for multiboot support.<br>
>> >> >> Unless<br>
>> >> >> that eliminates a LOT of PCs.<br>
>> >> >><br>
>> >> >> I don't want you to spend all summer getting an image to boot both<br>
>> >> >> ways. Personally, I want you to have a working BSP one way. :)<br>
>> >> > +1<br>
>> >> ><br>
>> >><br>
>> >> Noted, thanks!<br>
>> >><br>
>> >> >>><br>
>> >> >>><br>
>> >> >>> [1] <a href="https://www.sourceware.org/ml/binutils/2018-05/msg00197.html" rel="noreferrer noreferrer" target="_blank">https://www.sourceware.org/ml/binutils/2018-05/msg00197.html</a><br>
>> >> >>> [2] <a href="https://sourceforge.net/projects/gnu-efi/" rel="noreferrer noreferrer" target="_blank">https://sourceforge.net/projects/gnu-efi/</a><br>
>> >> >><br>
>> >> >><br>
>> >> >> --joel<br>
>> >> >><br>
>> >> >> _______________________________________________<br>
>> >> >> devel mailing list<br>
>> >> >> <a href="mailto:devel@rtems.org" target="_blank" rel="noreferrer">devel@rtems.org</a><br>
>> >> >> <a href="http://lists.rtems.org/mailman/listinfo/devel" rel="noreferrer noreferrer" target="_blank">http://lists.rtems.org/mailman/listinfo/devel</a><br>
</blockquote></div></div></div>