Commit Graph

649 Commits

Author SHA1 Message Date
John Baldwin
2a7278307b Remove a duplicate store when performing REL32 relocations in rtld.
The duplicate call to store_ptr() was added in r204687, but it should
have no effect as it only stores an Elf_Sword and the later store_ptr()
does a write that is at least as large if not larger.

Reviewed by:	jmallett
Obtained from:	CheriBSD (sort of)
Sponsored by:	DARPA / AFRL
2017-01-30 23:13:41 +00:00
Peter Jeremy
8787928589 Extend LD_UTRACE by also generating utrace(2) log events for runtime linker
errors.

Reviewed by:	kib, jhb
Approved by:	jhb(mentor)
MFC after:	1 week
Differential Revision:	 D9347
2017-01-30 08:38:32 +00:00
Andrew Turner
02dbdb1677 Pull the R_AARCH64_TLSDESC code out into a common function and use them in
both the plt and non-plt case.

This fixes an issue where libraries built with LLD can fail with
"Unhandled relocation 1031"

PR:		214971
Obtained from:	1 week
Sponsored by:	DARPA, AFRL
2017-01-25 17:35:11 +00:00
Ed Maste
ebf8934652 rtld: do not rely on a populated GOT on amd64
On rela architectures GNU BFD ld and gold store the relocation addend
in GOT entries (in addition to the relocation's r_addend field).
rtld previously relied on this to access its own _DYNAMIC symbol in
order to apply its own relocations.

However, recording addends in the GOT is not specified by the ABI,
and some versions of LLVM's LLD linker leave the GOT uninitialized on
rela architectures.

BFD ld does not populate the GOT on sparc64, and sparc64 rtld has a
machine-dependent rtld_dynamic_addr() function that returns the
_DYNAMIC address. Use the same approach on amd64, obtaining the %rip-
relative _DYNAMIC address following a suggestion from Rafael Espíndola.

Architectures other than amd64 should be addressed in future work.

PR:		214972
Reviewed by:	kib
MFC after:	2 weeks
Sponsored by:	The FreeBSD Foundation
Differential Revision:	https://reviews.freebsd.org/D9180
2017-01-16 14:49:29 +00:00
Konstantin Belousov
f8adf1a784 For the main binary, postpone enforcing relro read-only protection
until copy relocations are done.

Newer binutils and lld seems to output copy into relro-protected range.

Reported by: Rafael Espц╜ndola via emaste
Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
2017-01-12 15:54:03 +00:00
Konstantin Belousov
b88a8d3d1d Fix acquisition of nested write compat rtld locks.
Obtaining compat rtld lock in write mode sets process signal mask to
block all signals.  Previous mask is stored in the global variable
oldsigmask.  If a lock is write-locked while another lock is already
write-locked, oldsigmask is overwritten by the total mask and on the
last unlock, all signals except traps appear to be blocked.

Fix this by counting the write-lock nested level, and only storing to
oldsigmask/restoring from it at the outermost level.

Masking signals disables involuntary preemption for libc_r, and there
could be no voluntary context switches in the locked code
(dl_iterate_phdr(3) keeps a lock around user callback, but it was
added long after libc_r was renounced).  Due to this, remembering the
level in the global variable after the lock is obtained should be
safe, because no two libc_r threads can acquire different write locks
in parallel.

PR:	215826
Reported by:	kami
Tested by:	yamagi@yamagi.org (previous version)
To be reviewed by:	kan
Sponsored by:	The FreeBSD Foundation
MFC after:	2 weeks
2017-01-10 19:26:55 +00:00
Konstantin Belousov
e7bfd34bdf Use ANSI C definitions, update comment.
Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
2017-01-10 17:05:34 +00:00
Mark Johnston
57a9273f93 rtld: Fix a couple of bugs around the unloading of ELF filters.
- Pass the correct object to unload_filtees().
- Use a marker to restart iteration after unload_filtees() has returned.
  It calls dlclose() and may recursively remove entries from the global
  object list, so TAILQ_FOREACH_SAFE is not sufficient.

Reviewed by:	kib
MFC after:	2 weeks
Sponsored by:	Dell EMC Isilon
2016-12-22 17:44:27 +00:00
Mark Johnston
510fe58c82 rtld: Ensure that dlopen() cannot obtain a reference on a doomed object.
rtld drops the bind lock to call fini functions in an object prior to
unmapping it. The new "doomed" state flag prevents the acquisition of new
references for an object while the lock is dropped.

Reviewed by:	kib
MFC after:	2 weeks
Sponsored by:	Dell EMC Isilon
2016-12-22 17:41:32 +00:00
Mark Johnston
c02741759f rtld: Fix a race between dl_iterate_phdr() and dlclose().
Add a transient reference count to ensure that the phdr argument to the
callback remains valid while the bind lock is dropped.

Reviewed by:	kib
MFC after:	2 weeks
Sponsored by:	Dell EMC Isilon
2016-12-22 17:37:39 +00:00
Konstantin Belousov
3931b59fad Fix typo.
MFC after:	3 days
2016-12-16 14:23:08 +00:00
Bryan Drewery
9d9b69b373 Take write lock for rtld_bind before modifying obj_list in dl_iterate_phdr().
This avoids a race with readers such as dladdr(3)/dlinfo(3)/dlsym(3) and
the atexit(3) handler.  This race was introduced in r294373.

Reviewed by:	markj, kib, kan
MFC after:	2 weeks
Sponsored by:	Dell EMC Isilon
2016-12-13 18:05:14 +00:00
Ed Maste
eeef183376 Retire long-broken/unused static rtld support
rtld-elf has some vestigial support for building as a static executable.
r45501 introduced a partial implementation with a prescient note that it
"might never be enabled." r153515 introduced ELF symbol versioning
support, and removed part of the unused build infrastructure for static
rtld.

GNU ld populates rela relocation addends and GOT entries with the same
values, and rtld's run-time dynamic executable check relied on this.
Alternate toolchains may not populate the GOT entries, which caused
RTLD_IS_DYNAMIC to return false. Simplify rtld by just removing the
unused check.

If we want to restore static rtld support later on we ought to introduce
a build-time #ifdef flag.

PR:		214972
Reviewed by:	kan
MFC after:	1 month
Sponsored by:	The FreeBSD Foundation
Differential Revision:	https://reviews.freebsd.org/D8687
2016-12-02 14:23:26 +00:00
John Baldwin
6a368d16c8 Fix _mips_rtld_bind() to handle ELF filters.
MIPS does not use the common _rtld_bind() to handle runtime binding.
Instead, it uses a private _mips_rtld_bind().  Update _mips_rtld_bind()
to include the changes made to _rtld_bind() in r216695 and r218476 to
support upgrading the read-locked rtld_bind_lock to a write lock when
an object with a filter is encountered.

While here, add a 'where' variable to track the location of the fixup
in the GOT to make the code flow more closely match _rtld_bind().

Reviewed by:	kib
Obtained from:	CheriBSD
Sponsored by:	DARPA / AFRL
Differential Revision:	https://reviews.freebsd.org/D8625
2016-11-23 20:21:53 +00:00
Konstantin Belousov
d27078f990 Adjust r308689 to make rtld compilable with either in-tree or
(hopefully) stock gcc 4.2.1 on i386 and other arches.

In particular:
- Do not use %ebx in the asm constraints on i386, since rtld is
  compiled with -fPIC and gcc cannot handle GOT-base register reload
  (clang and newer gcc can).
- Avoid direct use of [static N] construct in the function
  declaration/definion.  In-tree gcc was patched to support this, but
  stock 4.2.1 cannot handle the feature.

Requested by:	bde
Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
2016-11-21 14:13:57 +00:00
Konstantin Belousov
4352999e0e Pass CPUID[1] %edx (cpu_feature), %ecx (cpu_feature2) and
CPUID[7].%ebx (cpu_stdext_feature), %ecx (cpu_stdext_feature2) to the
ifunc resolvers on x86.

It is much more clean to use CPUID instruction in usermode to retrieve
this information than to pass AT_HWCAP aux vector from kernel, on
x86.  Still, the change does allow for use of AT_HWCAP on arches where it is
needed, by passing aux array to ifunc_init() initializer which should
prepare arguments for ifunc resolvers.

Current signature for resolvers on x86 is
	func_t iresolve(uint32_t cpu_feature, uint32_t cpu_feature2,
	    uint32_t cpu_stdext_feature, uint32_t cpu_stdext_feature2);
where arguments have identical meaning as the kernel variables of the
same name.  The ABIs allow to use resolvers with the void or shortened
list of arguments.

Reviewed by:	jhb
Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
Differential revision:	https://reviews.freebsd.org/D8448
2016-11-15 09:43:26 +00:00
Konstantin Belousov
4562cfc40e Assert that there is no unresolved symbols during rtld linking.
Reviewed by:	emaste, jhb
Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
X-Differential revision:	https://reviews.freebsd.org/D8448
2016-11-15 09:40:03 +00:00
Konstantin Belousov
093513c76b Update hint to utilize user variable.
Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
2016-11-15 09:37:35 +00:00
John Baldwin
b5b4f379e0 Pass the correct flag to find_symdef() from _rtld_bind().
When symbol versioning was added to rtld, the boolean 'in_plt' argument
to find_symdef() was converted to a bitmask of flags.  The first flag
added was 'SYMLOOK_IN_PLT' which replaced the 'in_plt' bool.  This
happened to still work by accident as SYMLOOK_IN_PLT had the value of 1
which is the same as 'true', so there should be no functional change.

Tested on:	amd64
Reviewed by:	kan
MFC after:	2 weeks
Sponsored by:	DARPA / AFRL
2016-11-08 22:41:11 +00:00
Bryan Drewery
1eb4ec9c62 DIRDEPS_BUILD: Add some missing dirctories to the build.
Sponsored by:	EMC / Isilon Storage Division
2016-08-31 19:30:59 +00:00
Adrian Chadd
6f9dab906e [mips] flip from =v to =r - let the compiler choose a temp register.
=v is some ye olde gcc "use this specific register as the temp register"
thing that they've deprecated and clang/llvm doesn't implement.

Poked again and again by: sbruno
2016-08-25 22:32:10 +00:00
Konstantin Belousov
f214036e99 Move defines common between rtld and libsysdecode into the header,
instead of copying inline into sources.

Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
2016-08-12 19:31:41 +00:00
Konstantin Belousov
09944776b5 Fill phdr and phsize for rtld object. It is needed for
dl_iterate_phdr() reporting the correct values.

PR:	211367
Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
2016-08-12 18:31:44 +00:00
Konstantin Belousov
20ee0f7112 Remove all remaining uses of TAILQ_FOREACH_FROM() from rtld-elf.
Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
2016-08-12 18:29:11 +00:00
Adrian Chadd
b812fe4d6b [mips] add support for using the MIPS user register for TLS data.
This work, originally from Stacey Son, uses the MIPS UserReg for
reading the TLS data, and will fall back to the normal syscall path
when it isn't supported.

This code dynamically patches cpu_switch() to bypass the UserReg
instruction so to avoid generating a machine exception.

Thanks to sson for the original work, and to Dan Nelson for
bringing it to date and testing it on MIPS32 with me.

Tested:

* mips64 (sson)
* mips74k (dnelson_1901@yahoo.com) - AR9344 SoC, UserReg support
* mips24k (adrian) - AR9331 SoC, no UserReg support

Obtained from:	sson, dnelson_1901@yahoo.com
2016-08-07 01:29:55 +00:00
Bryan Drewery
1c1093d6d6 Fix dlsym(RTLD_NEXT) handling to only return the next library in last library cases.
The root of the problem here is that TAILQ_FOREACH_FROM will default to
the head of the list if passed NULL, which will be the case if there are
no libraries loaded after this one.  Thus all libraries, including the
current, were iterated in that case rather than none.

This was broken in r294373.

Reviewed by:	markj (earlier version), cem, kib, ngie
MFC after:	1 week
Sponsored by:	EMC / Isilon Storage Division
Differential Revision:	https://reviews.freebsd.org/D7216
2016-07-15 19:07:00 +00:00
Bryan Drewery
e7debdcb15 Fix race for incrementally rebuilding VERSION_MAP.
The dependency is needed in PROG_FULL since only the build of PROG_FULL
is using the LDFLAGS and depending on VERSION_MAP.  This was not a problem
with MK_DEBUG_FILES==no since it only builds PROG.

This should probably be using bsd.lib.mk instead [1]

Reported by:	swills, gjb
Reviewed by:	emaste
Noted by:	rgrimes [1]
Sponsored by:	EMC / Isilon Storage Division
Approved by:	re (kib)
2016-07-03 17:28:39 +00:00
Konstantin Belousov
360ba6bc13 Fix issues found by Coverity in the rtld-elf.c:gethints().
Check that the dirlist path string specification does not cause
overflow and is fully contained in the hints file.
Check that the dirlist string is nul-terminated.
Make 'hdr' static variable, so that hdr.dirlistlen is available when
hints cached value is used on next function calls.  Reset hdr.dirlistlen
to zero if error was detected, so that allocations use reasonable size.
Use 'hints', and not 'p' in the body, since p is only initialized on the
first call.

Reported and reviewed by:	truckman (previous version)
Sponsored by:	The FreeBSD Foundation
CIDs:	1006503, 1006504, 1006676, 1008488, 1007263
MFC after:	2 weeks
2016-05-25 18:10:44 +00:00
Enji Cooper
430f7286a5 Merge ^/user/ngie/release-pkg-fix-tests to unbreak how test files are installed
after r298107

Summary of changes:

- Replace all instances of FILES/TESTS with ${PACKAGE}FILES. This ensures that
  namespacing is kept with FILES appropriately, and that this shouldn't need
  to be repeated if the namespace changes -- only the definition of PACKAGE
  needs to be changed
- Allow PACKAGE to be overridden by callers instead of forcing it to always be
  `tests`. In the event we get to the point where things can be split up
  enough in the base system, it would make more sense to group the tests
  with the blocks they're a part of, e.g. byacc with byacc-tests, etc
- Remove PACKAGE definitions where possible, i.e. where FILES wasn't used
  previously.
- Remove unnecessary TESTSPACKAGE definitions; this has been elided into
  bsd.tests.mk
- Remove unnecessary BINDIRs used previously with ${PACKAGE}FILES;
  ${PACKAGE}FILESDIR is now automatically defined in bsd.test.mk.
- Fix installation of files under data/ subdirectories in lib/libc/tests/hash
  and lib/libc/tests/net/getaddrinfo
- Remove unnecessary .include <bsd.own.mk>s (some opportunistic cleanup)

Document the proposed changes in share/examples/tests/tests/... via examples
so it's clear that ${PACKAGES}FILES is the suggested way forward in terms of
replacing FILES. share/mk/bsd.README didn't seem like the appropriate method
of communicating that info.

MFC after: never probably
X-MFC with: r298107
PR: 209114
Relnotes: yes
Tested with: buildworld, installworld, checkworld; buildworld, packageworld
Sponsored by: EMC / Isilon Storage Division
2016-05-04 23:20:53 +00:00
Pedro F. Giffuni
ec489d64bb libexec: minor spelling fixes in comments.
No functional change.
2016-05-01 19:39:23 +00:00
Pedro F. Giffuni
260111ffce rtld-elf: use our roundup2() macro when param.h is available. 2016-04-19 20:12:46 +00:00
Pedro F. Giffuni
a5d5e8dd7c rtld-elf: use NULL instead of zero for pointers. 2016-04-19 19:03:55 +00:00
Glen Barber
d60840138f MFH
Sponsored by:	The FreeBSD Foundation
2016-04-04 23:55:32 +00:00
Bryan Drewery
c0f5aeb032 WITHOUT_TOOLCHAIN: Fix build of rtld.
MK_TOOLCHAIN==no disables building and installing of pic archives.
c_pic.a is still needed for rtld though so force it to build in lib/libc
and link directly to the objdir version of it for rtld.

Somehow this has been broken since r148725.

Sponsored by:	EMC / Isilon Storage Division
2016-03-31 17:27:01 +00:00
Glen Barber
7d536dc855 MFH
Sponsored by:	The FreeBSD Foundation
2016-03-10 21:16:01 +00:00
Bryan Drewery
15c433351f DIRDEPS_BUILD: Connect MK_TESTS.
Sponsored by:	EMC / Isilon Storage Division
2016-03-09 22:46:01 +00:00
Glen Barber
42d27ee343 MFH
Sponsored by:	The FreeBSD Foundation
2016-03-02 23:53:08 +00:00
Konstantin Belousov
ca8c8dc3eb Fix handling of DT_TEXTREL for an object with more than one read-only
segment.  According to gABI spec, presence of the tag indicates that
dynamic linker must be prepared to handle relocations against any
read-only segment, not only the segment which we, somewhat arbitrary,
declared the text.

For each read-only segment, add write permission before relocs are
processed, and return to the mapping mode requested by the phdr, after
relocs are done.

Reported, tested, and reviewed by:	emaste
PR:	207631
Sponsored by:	The FreeBSD Foundation
MFC after:	2 weeks
2016-03-02 16:36:24 +00:00
Glen Barber
52259a98ad MFH
Sponsored by:	The FreeBSD Foundation
2016-03-02 16:14:46 +00:00
Mariusz Zaborski
c501d73c7e Convert casperd(8) daemon to the libcasper.
After calling the cap_init(3) function Casper will fork from it's original
process, using pdfork(2). Forking from a process has a lot of advantages:
1. We have the same cwd as the original process.
2. The same uid, gid and groups.
3. The same MAC labels.
4. The same descriptor table.
5. The same routing table.
6. The same umask.
7. The same cpuset(1).
From now services are also in form of libraries.
We also removed libcapsicum at all and converts existing program using Casper
to new architecture.

Discussed with:		pjd, jonathan, ed, drysdale@google.com, emaste
Partially reviewed by:	drysdale@google.com, bdrewery
Approved by:		pjd (mentor)
Differential Revision:	https://reviews.freebsd.org/D4277
2016-02-25 18:23:40 +00:00
Glen Barber
72c3aa02dc MFH
Sponsored by:	The FreeBSD Foundation
2016-02-18 00:37:58 +00:00
Bryan Drewery
fb99179af4 Fix build race after r295643.
Sponsored by:	EMC / Isilon Storage Division
2016-02-17 00:30:28 +00:00
Glen Barber
c389411c8a Remove libc, librtld_db, libthr packages, and further increase
the constraints on what needs to be installed in a specific to
maintain consistency during upgrades.

Create a new clibs package containing libraries that are needed
as a bare minimum for consistency.

With much help and input from:	kib
Sponsored by:	The FreeBSD Foundation
2016-02-05 21:01:08 +00:00
Glen Barber
393303e317 Include ld-elf.so and ld-elf32.so in the librtld_db
package.

Sponsored by:	The FreeBSD Foundation
2016-02-05 02:29:41 +00:00
Glen Barber
221b349912 MFH
Sponsored by:	The FreeBSD Foundation
2016-02-02 22:27:48 +00:00
Glen Barber
43faedc133 First pass to fix the 'tests' packages.
Sponsored by:	The FreeBSD Foundation
2016-02-02 22:26:49 +00:00
Maxim Sobolev
7fd852f860 This seems like a very trivial bug that should have been squashed a long
time ago, but for some reason it was not. Basically, without this change
dlopen(3)'ing an empty .so file would just cause application to dump core
with SIGSEGV.

Make sure the file has enough data for at least the ELF header before
mmap'ing it.

Add a test case to check that dlopen an empty file return an error.

There were a separate discussion as to whether it should be SIGBUS
instead when you try to access region mapped from an empty file,
but it's definitely SIGSEGV now, so if anyone want to check that please
be my guest.
Reviewed by:	mjg, cem
MFC after:	1 week
Differential Revision:	https://reviews.freebsd.org/D5112
2016-01-30 04:16:05 +00:00
Alexander Kabaev
ef2c2a692b Do not unlock rtld_phdr_lock over callback invocations.
The dl_iterate_phdr consumer code in libgcc does not expect multiple
callbacks running concurrently. This was fixed once already in r178807,
but accidentally got reverted in r294373.
2016-01-27 20:20:37 +00:00
Ruslan Bukin
e07528700a Follow r293066 adding a generalized exec hook for RISC-V as well. 2016-01-23 11:46:52 +00:00
Alexander Kabaev
832b947305 Fix initlist_add_object invocation parameters.
The tail parameter should point to the last object for
which dependencies should be processed. In most cases,
this is the object itself.
2016-01-20 23:26:35 +00:00