After our migration (of certain architectures) to lld the kernel is built
with a unique build-ID. Make it available via a sysctl and uname(1) to
allow the user to identify their running kernel.
Submitted by: Ali Mashtizadeh <ali_mashtizadeh.com>
MFC after: 2 weeks
Relnotes: Yes
Event: Waterloo Hackathon 2019
Differential Revision: https://reviews.freebsd.org/D20326
With lld 7.0.0, a rather nasty problem in our kernel linker script came
to light. We use quite a lot of so-called "orphan" sections, e.g.
sections which are not explicitly named in the linker script. Mainly,
these are the linker sets (such as set_sysinit_set).
Note that the placement of these orphan sections is not very well
defined. Usually, any read-only orphan sections get placed after the
last read-only section from the linker script, and similarly for the
read/write variants.
In our linker scripts, there are also symbol assignments like _etext,
_edata, and __bss_start, which are used in various places to refer to
the start or end addresses of sections.
However, some of these symbol assignments are interspersed with output
section descriptions. While the linker will guarantee that a symbol
assignment after some section will stay after that section, there is no
guarantee that an orphan section cannot be inserted just before it.
Take for example the following script:
SECTIONS
{
.data : { *(.data) }
__bss_start = .;
.bss : { *(.bss) }
}
If an orphan section (like set_sysinit_set) is now inserted just after
the __bss_start assignment, __bss_start will actually point to the start
of that orphan section, *not* to the start of the .bss section.
Unfortunately, something like this happened with our i386 kernel linker
script, and since sys/i386/i386/locore.s tries to zero .bss, it ended up
zeroing all the linker sets too, leading to a crash very soon after the
<--BOOT--> message.
To fix this, move the __bss_start symbol assignment *into* the .bss
section description, so there is no way a linker can then insert orphan
sections at that point. Also add a corresponding __bss_end symbol.
In addition, change sys/i386/i386/locore.s, so it clears from
__bss_start to __bss_end, instead of assuming that _edata is just
before .bss (which may not be true), and that _end is just after _bss
(which also may not be true).
This allows an i386 kernel linked with lld 7.0.0 to boot successfully.
The change makes the user and kernel address spaces on i386
independent, giving each almost the full 4G of usable virtual addresses
except for one PDE at top used for trampoline and per-CPU trampoline
stacks, and system structures that must be always mapped, namely IDT,
GDT, common TSS and LDT, and process-private TSS and LDT if allocated.
By using 1:1 mapping for the kernel text and data, it appeared
possible to eliminate assembler part of the locore.S which bootstraps
initial page table and KPTmap. The code is rewritten in C and moved
into the pmap_cold(). The comment in vmparam.h explains the KVA
layout.
There is no PCID mechanism available in protected mode, so each
kernel/user switch forth and back completely flushes the TLB, except
for the trampoline PTD region. The TLB invalidations for userspace
becomes trivial, because IPI handlers switch page tables. On the other
hand, context switches no longer need to reload %cr3.
copyout(9) was rewritten to use vm_fault_quick_hold(). An issue for
new copyout(9) is compatibility with wiring user buffers around sysctl
handlers. This explains two kind of locks for copyout ptes and
accounting of the vslock() calls. The vm_fault_quick_hold() AKA slow
path, is only tried after the 'fast path' failed, which temporary
changes mapping to the userspace and copies the data to/from small
per-cpu buffer in the trampoline. If a page fault occurs during the
copy, it is short-circuit by exception.s to not even reach C code.
The change was motivated by the need to implement the Meltdown
mitigation, but instead of KPTI the full split is done. The i386
architecture already shows the sizing problems, in particular, it is
impossible to link clang and lld with debugging. I expect that the
issues due to the virtual address space limits would only exaggerate
and the split gives more liveness to the platform.
Tested by: pho
Discussed with: bde
Sponsored by: The FreeBSD Foundation
MFC after: 1 month
Differential revision: https://reviews.freebsd.org/D14633
We should never end up executing the inter-function padding, so we
are better off faulting than silently carrying on to whatever function
happens to be next.
Note that LLD will soon do this by default (although it currently pads
with zeros).
Reviewed by: dim, kib
MFC after: 1 month
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D10047
The linker script CONSTRUCTORS keyword is only meaningful "when linking
object file formats which do not support arbitrary sections, such as
ECOFF and XCOFF"[1] and is ignored for other object file formats.
LLVM's lld does not yet accept (and ignore) CONSTRUCTORS, so just remove
CONSTRUCTORS from the linker scripts as it has no effect.
[1] https://sourceware.org/binutils/docs/ld/Output-Section-Keywords.html
Reviewed by: kib
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D7343
avoid problems with some Pentium 4 cpus and some older PPro/Pentium2
cpus. There are several problems, some documented in Intel errata.
This patch:
1) moves the kernel to the second page in the PSE case. There is an
errata that says that you Must Not point a 4MB page at physical
address zero on older cpus. We avoided bugs here due to sheer luck.
2) sets up PSE page tables right from the start in locore, rather than
trying to switch from 4K to 4M (or 2M) pages part way through the boot
sequence at the same time that we're messing with PG_G.
For some reason, the pmap work over the last 18 months seems to tickle
the problems, and the PAE infrastructure changes disturb the cpu
bugs even more.
A couple of people have reported a problem with APM bios calls during
boot. I'll work with people to get this resolved.
Obtained from: bmilekic
bootblocks in order to boot the kernel after this! Also note that this
change breaks BSDI BSD/OS compatibility.
Also increased default NKPT to 17 so that FreeBSD can boot on machines
with >=2GB of RAM. Booting on machines with exactly 4GB requires other
patches, not included.
the in-kernel linker to access the _DYNAMIC data for doing loadable elf
modules. The alpha kernel is already done this way, I've borrowed some of
the hacks from there.
This is primarily aimed at the 3-stage boot process which is intended to
be able to do pre-loading of kernel modules.
Note that the entry point isn't 0xf0100000 any more, it'll be a little
further on - but this value is stored in the headers. I don't think this
will be a problem, but I'm sure somebody will tell me if it is. :-)
I'm not sure if btxboot is going to like this, it doesn't do proper ELF
header checking and assumes that there are exactly two program header
entries and that they are both PT_LOAD entries - a bad assumption.
