Files
operating-system/buildroot/package/patchelf/0005-Avoid-inflating-file-sizes-needlessly-and-allow-bina.patch
Stefan Agner a0871be6c0 Bump buildroot to 2020.11-rc1 (#985)
* Update buildroot-patches for 2020.11-rc1 buildroot

* Update buildroot to 2020.11-rc1

Signed-off-by: Stefan Agner <stefan@agner.ch>

* Don't rely on sfdisk --list-free output

The --list-free (-F) argument does not allow machine readable mode. And
it seems that the output format changes over time (different spacing,
using size postfixes instead of raw blocks).

Use sfdisk json output and calculate free partition space ourselfs. This
works for 2.35 and 2.36 and is more robust since we rely on output which
is meant for scripts to parse.

* Migrate defconfigs for Buildroot 2020.11-rc1

In particular, rename BR2_TARGET_UBOOT_BOOT_SCRIPT(_SOURCE) to
BR2_PACKAGE_HOST_UBOOT_TOOLS_BOOT_SCRIPT(_SOURCE).

* Rebase/remove systemd patches for systemd 246

* Drop apparmor/libapparmor from buildroot-external

* hassos-persists: use /run as directory for lockfiles

The U-Boot tools use /var/lock by default which is not created any more
by systemd by default (it is under tmpfiles legacy.conf, which we no
longer install).

* Disable systemd-update-done.service

The service is not suited for pure read-only systems. In particular the
service needs to be able to write a file in /etc and /var. Remove the
service. Note: This is a static service and cannot be removed using
systemd-preset.

* Disable apparmor.service for now

The service loads all default profiles. Some might actually cause
problems. E.g. the profile for ping seems not to match our setup for
/etc/resolv.conf:
[85503.634653] audit: type=1400 audit(1605286002.684:236): apparmor="DENIED" operation="open" profile="ping" name="/run/resolv.conf" pid=27585 comm="ping" requested_mask="r" denied_mask="r" fsuid=0 ouid=0
2020-11-13 18:25:44 +01:00

177 lines
7.6 KiB
Diff
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
From 79c093226e609b99fa889f6e37480b92b399610d Mon Sep 17 00:00:00 2001
From: Richard Purdie <richard.purdie@linuxfoundation.org>
Date: Tue, 7 Mar 2017 21:08:34 +0000
Subject: [PATCH] Avoid inflating file sizes needlessly and allow binaries to
be stripped
The current approach to changing sections in ET_DYN executables is to move
the INTERP section to the end of the file. +This means changing PT_PHDR to
add an extra PT_LOAD section so that the new section is mmaped into memory
by the elf loader in the kernel. In order to extend PHDR, this means moving
it to the end of the file.
Its documented in BUGS there is a kernel 'bug' which means that if you have holes
in memory between the base load address and the PT_LOAD segment that contains PHDR,
it will pass an incorrect PHDR address to ld.so and fail to load the binary, segfaulting.
To avoid this, the code currently inserts space into the binary to ensure that when
loaded into memory there are no holes between the PT_LOAD sections. This inflates the
binaries by many MBs in some cases. Whilst we could make them sparse, there is a second
issue which is that strip can fail to process these binaries:
$ strip fixincl
Not enough room for program headers, try linking with -N
[.note.ABI-tag]: Bad value
This turns out to be due to libbfd not liking the relocated PHDR section either
(https://github.com/NixOS/patchelf/issues/10).
Instead this patch implements a different approach, leaving PHDR where it is but extending
it in place to allow addition of a new PT_LOAD section. This overwrites sections in the
binary but those get moved to the end of the file in the new PT_LOAD section.
This is based on patches linked from the above github issue, however whilst the idea
was good, the implementation wasn't correct and they've been rewritten here.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fetch from: https://github.com/NixOS/patchelf/commit/c4deb5e9e1ce9c98a48e0d5bb37d87739b8cfee4
Backported to v0.9
Signed-off-by: Conrad Ratschan <conrad.ratschan@rockwellcollins.com>
---
src/patchelf.cc | 71 ++++++++++++++++++++++++++++---------------------
1 file changed, 40 insertions(+), 31 deletions(-)
diff --git a/src/patchelf.cc b/src/patchelf.cc
index 1d58061..c2147af 100644
--- a/src/patchelf.cc
+++ b/src/patchelf.cc
@@ -209,6 +209,8 @@ private:
string & replaceSection(const SectionName & sectionName,
unsigned int size);
+ bool haveReplacedSection(const SectionName & sectionName);
+
void writeReplacedSections(Elf_Off & curOff,
Elf_Addr startAddr, Elf_Off startOffset);
@@ -632,6 +634,15 @@ void ElfFile<ElfFileParamNames>::writeReplacedSections(Elf_Off & curOff,
replacedSections.clear();
}
+template<ElfFileParams>
+bool ElfFile<ElfFileParamNames>::haveReplacedSection(const SectionName & sectionName)
+{
+ ReplacedSections::iterator i = replacedSections.find(sectionName);
+
+ if (i != replacedSections.end())
+ return true;
+ return false;
+}
template<ElfFileParams>
void ElfFile<ElfFileParamNames>::rewriteSectionsLibrary()
@@ -648,52 +659,53 @@ void ElfFile<ElfFileParamNames>::rewriteSectionsLibrary()
debug("last page is 0x%llx\n", (unsigned long long) startPage);
+ /* Because we're adding a new section header, we're necessarily increasing
+ the size of the program header table. This can cause the first section
+ to overlap the program header table in memory; we need to shift the first
+ few segments to someplace else. */
+ /* Some sections may already be replaced so account for that */
+ unsigned int i = 1;
+ Elf_Addr pht_size = sizeof(Elf_Ehdr) + (phdrs.size() + 1)*sizeof(Elf_Phdr);
+ while( shdrs[i].sh_addr <= pht_size && i < rdi(hdr->e_shnum) ) {
+ if (not haveReplacedSection(getSectionName(shdrs[i])))
+ replaceSection(getSectionName(shdrs[i]), shdrs[i].sh_size);
+ i++;
+ }
- /* Compute the total space needed for the replaced sections and
- the program headers. */
- off_t neededSpace = (phdrs.size() + 1) * sizeof(Elf_Phdr);
+ /* Compute the total space needed for the replaced sections */
+ off_t neededSpace = 0;
for (ReplacedSections::iterator i = replacedSections.begin();
i != replacedSections.end(); ++i)
neededSpace += roundUp(i->second.size(), sectionAlignment);
debug("needed space is %d\n", neededSpace);
-
size_t startOffset = roundUp(fileSize, getPageSize());
growFile(startOffset + neededSpace);
-
/* Even though this file is of type ET_DYN, it could actually be
an executable. For instance, Gold produces executables marked
- ET_DYN. In that case we can still hit the kernel bug that
- necessitated rewriteSectionsExecutable(). However, such
- executables also tend to start at virtual address 0, so
+ ET_DYN as does LD when linking with pie. If we move PT_PHDR, it
+ has to stay in the first PT_LOAD segment or any subsequent ones
+ if they're continuous in memory due to linux kernel constraints
+ (see BUGS). Since the end of the file would be after bss, we can't
+ move PHDR there, we therefore choose to leave PT_PHDR where it is but
+ move enough following sections such that we can add the extra PT_LOAD
+ section to it. This PT_LOAD segment ensures the sections at the end of
+ the file are mapped into memory for ld.so to process.
+ We can't use the approach in rewriteSectionsExecutable()
+ since DYN executables tend to start at virtual address 0, so
rewriteSectionsExecutable() won't work because it doesn't have
- any virtual address space to grow downwards into. As a
- workaround, make sure that the virtual address of our new
- PT_LOAD segment relative to the first PT_LOAD segment is equal
- to its offset; otherwise we hit the kernel bug. This may
- require creating a hole in the executable. The bigger the size
- of the uninitialised data segment, the bigger the hole. */
+ any virtual address space to grow downwards into. */
if (isExecutable) {
if (startOffset >= startPage) {
debug("shifting new PT_LOAD segment by %d bytes to work around a Linux kernel bug\n", startOffset - startPage);
- } else {
- size_t hole = startPage - startOffset;
- /* Print a warning, because the hole could be very big. */
- fprintf(stderr, "warning: working around a Linux kernel bug by creating a hole of %zu bytes in %s\n", hole, fileName.c_str());
- assert(hole % getPageSize() == 0);
- /* !!! We could create an actual hole in the file here,
- but it's probably not worth the effort. */
- growFile(fileSize + hole);
- startOffset += hole;
}
startPage = startOffset;
}
- /* Add a segment that maps the replaced sections and program
- headers into memory. */
+ /* Add a segment that maps the replaced sections into memory. */
phdrs.resize(rdi(hdr->e_phnum) + 1);
wri(hdr->e_phnum, rdi(hdr->e_phnum) + 1);
Elf_Phdr & phdr = phdrs[rdi(hdr->e_phnum) - 1];
@@ -706,15 +718,12 @@ void ElfFile<ElfFileParamNames>::rewriteSectionsLibrary()
/* Write out the replaced sections. */
- Elf_Off curOff = startOffset + phdrs.size() * sizeof(Elf_Phdr);
+ Elf_Off curOff = startOffset;
writeReplacedSections(curOff, startPage, startOffset);
assert(curOff == startOffset + neededSpace);
-
- /* Move the program header to the start of the new area. */
- wri(hdr->e_phoff, startOffset);
-
- rewriteHeaders(startPage);
+ /* Write out the updated program and section headers */
+ rewriteHeaders(hdr->e_phoff);
}
--
2.17.1