Change 935f199875 ('create-diff-object: simplify mangled function
correlation') simplified the way symbols are correlated and got rid of
symbol section renaming.
As a result a symbol/section can now have a CHANGED status, being
correlated to an element that doesn't have the exact same name. This
will cause lookups to the original object fail when creating the new
patch object.
So lets bring back the symbol/section renaming, but only once they
have actually been correlated.
Fixes: 935f199875 ('create-diff-object: simplify mangled function
correlation')
Signed-off-by: Julien Thierry <jthierry@redhat.com>
Elements from the original object and the patched object can be
correlated using their mangled names. In case an elements (section or
symbol) could be matched with more than one object through mangling,
make sure all elements related to a section are correlated with the
corresponding elements of the twin section.
Signed-off-by: Julien Thierry <jthierry@redhat.com>
The RHEL powerpc kernel is compiled with -O3, which triggers some
"interesting" new optimizations. One of them, which seems to be
relatively common, is the replacing of a function with two separate
"constprop" functions.
Previously we only ever saw a single constprop clone, so we just renamed
the patched version of the function to match the original version. Now
that we can have two such clones, that no longer makes sense.
Instead of renaming functions, just improve the correlation logic such
that they can be correlated despite having slightly different symbol
names. The first clone in the original object is correlated with the
first clone in the patched object; the second clone is correlated with
the second clone; and so on.
This assumes that the order of the symbols and sections doesn't change,
which seems to be a reasonable assumption based on past experience with
the compiler. Otherwise it will just unnecessarily mark the cloned
constprop functions as changed, which is annoying but harmless, and
noticeable by a human anyway.
Fixes#935.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
If the symbol associated with a relocation does not have a section set,
nothing is done for that relocation.
Skip iterating through all the symbols of the ELF file in such a case.
Signed-off-by: Julien Thierry <jthierry@redhat.com>
rela_insn() only retrieves information about an instruction and does not
modify sections or relocations.
Add const to make this explicit.
Signed-off-by: Julien Thierry <jthierry@redhat.com>
There is no point inspecting through the symbols of the ELF files
(original and patched) when the ELF headers do not meet requirements.
Check ELF headers as soon as the files are mapped.
Signed-off-by: Julien Thierry <jthierry@redhat.com>
We are seeing the following error on a real world patch:
unsupported reference to special section __barrier_nospec_fixup
The kpatch commit bb444c2168 ("create-diff-object: Check for *_fixup
sections changes") created this error because we were trying to be
future proof. However, that may have been overly paranoid, as it
doesn't seem likely that those fixup sections will need relocations
anytime soon, because the replacement instructions are manually
generated in code. And anyway that "future proof" commit breaks the
present.
Also we decided at LPC that we are going to remove .klp.arch sections
anyway, so once that happens we will be fully future-proof anyway.
This reverts commit bb444c2168.
Fixes#974.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
While static keys (jump labels) are currently broken in livepatch, a
broken dynamic debug static key is harmless since it just disables
dynamically enabled debug printks in the patched code.
Fixes: #1021
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
We saw the following panic on ppc64le when loading the macro-callbacks
integration test:
livepatch: enabling patch 'kpatch_macro_callbacks'
Oops: Exception in kernel mode, sig: 4 [#1]
LE SMP NR_CPUS=2048 NUMA pSeries
Modules linked in: kpatch_macro_callbacks(OEK+) rpcsec_gss_krb5 auth_rpcgss nfsv4 dns_resolver nfs lockd grace fscache sunrpc sg pseries_rng xts vmx_crypto xfs libcrc32c sd_mod ibmvscsi scsi_transport_srp ibmveth dm_mirror dm_region_hash dm_log dm_mod [last unloaded: kpatch_gcc_static_local_var_6]
CPU: 2 PID: 17445 Comm: insmod Kdump: loaded Tainted: G OE K --------- - - 4.18.0-128.el8.ppc64le #1
NIP: d00000000bb708e0 LR: c0000000001fd610 CTR: d00000000bb708e0
REGS: c00000040e98f640 TRAP: 0700 Tainted: G OE K --------- - - (4.18.0-128.el8.ppc64le)
MSR: 800000000288b033 <SF,VEC,VSX,EE,FP,ME,IR,DR,RI,LE> CR: 28008228 XER: 20040003
CFAR: c0000000001fd60c IRQMASK: 0
GPR00: c0000000001fd5c0 c00000040e98f8c0 c000000001662a00 c000000733525400
GPR04: 0000000000000800 0000000000000800 c0000000015e2c00 c0000007335254a8
GPR08: 0000000000000001 d00000000bb708e0 c0000007eeb68400 0000000000000000
GPR12: d00000000bb708e0 c000000007fad600 0000000000000001 aaaaaaaaaaaaaaab
GPR16: 000000000000ff20 000000000000fff1 000000000000fff2 d00000000bb90000
GPR20: 00000000000000a9 c00000040e98fc00 c000000000d8a728 c00000040e98fc00
GPR24: d00000000bb73f88 00000000006080c0 d00000000bb73a38 c000000733525400
GPR28: 0000000000000001 c000000733525400 ffffffffffffffed c0000007eeb60900
NIP [d00000000bb708e0] callback_info.isra.0+0x7c/0x66c [kpatch_macro_callbacks]
LR [c0000000001fd610] __klp_enable_patch+0x130/0x230
Call Trace:
[c00000040e98f8c0] [c0000000001fd5c0] __klp_enable_patch+0xe0/0x230 (unreliable)
[c00000040e98f940] [c0000000001fd7d8] klp_enable_patch+0xc8/0x100
[c00000040e98f980] [d00000000bb7079c] patch_init+0x460/0x4cc [kpatch_macro_callbacks]
[c00000040e98fa20] [c000000000010108] do_one_initcall+0x58/0x248
[c00000040e98fae0] [c00000000023b860] do_init_module+0x80/0x330
[c00000040e98fb70] [c0000000002416a4] load_module+0x3994/0x3d00
[c00000040e98fd30] [c000000000241cf4] sys_finit_module+0xc4/0x130
[c00000040e98fe30] [c00000000000b388] system_call+0x5c/0x70
Instruction dump:
7cea482a 48000235 e8410018 48000014 3c620000 e8638160 48000221 e8410018
38210060 e8010010 7c0803a6 4e800020 <0000ae18> 00000000 3c4c0001 3842ae18
The problem was introduced by a recent fix:
e8f7f2dfe8 ("create-diff-object/ppc64le: Fix replace_sections_syms() for bundled symbols")
We didn't notice the fact that there's a hack in
kpatch_include_callback_elements() which reverts the work of
kpatch_replace_sections_syms() for callback function symbols.
The problem is that that revert is only partial, causing the callback
pointers to point to the .TOC data which is located 8 bytes before the
start of the function code. This happens because
kpatch_include_callback_elements() makes the same assumption that
kpatch_replace_sections_syms() had previously made: that bundled symbols
are always located at the start of their corresponding sections.
kpatch_include_callback_elements() mysteriously strips references to the
callback function symbols, replacing them with section symbols. In this
case it replaced a 'pre_patch_callback' function reference with a
'.text.unlikely.pre_patch_callback' section reference. But it didn't
adjust the rela->addend accordingly.
Joe discovered the reasoning for why kpatch_include_callback_elements()
removes function symbol references in the commit log for 7dfad2fb76
("fix dynrela corruption in load/unload hooks"):
In the case of the hook functions, we strip the FUNC symbol to prevent
it from being added to the kpatch.funcs section as a patched function.
But that justification doesn't really make sense, at least not with the
current code. Callbacks aren't added to .kpatch.funcs anyway. They're
classifed as NEW. Only CHANGED functions are added to .kpatch.funcs.
So remove that hack, fixing this bug in the process.
This does have a side effect of showing the callback functions as new
functions, because their symbols are now included.
Before:
aio.o: found callback: post_unpatch_callback
aio.o: found callback: pre_patch_callback
aio.o: found callback: pre_unpatch_callback
aio.o: new function: callback_info.isra.0
After:
aio.o: found callback: post_unpatch_callback
aio.o: found callback: pre_patch_callback
aio.o: found callback: pre_unpatch_callback
aio.o: new function: callback_info.isra.0
aio.o: new function: pre_patch_callback
aio.o: new function: post_patch_callback
aio.o: new function: pre_unpatch_callback
aio.o: new function: post_unpatch_callback
But anyway they _are_ new functions, so the new output seems more
correct to me.
Fixes: e8f7f2dfe8 ("create-diff-object/ppc64le: Fix replace_sections_syms() for bundled symbols")
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
The existing comment is wrong. It confusingly conflates the function's
offset, which is 8 bytes from the beginning of the section, with the
function's localentry offset which is 8 bytes from the beginning of the
function.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
group_size variable is assigned right after we enter for loop without
ever being read so there is no need to initialize it to 0 beforehand.
Signed-off-by: Artem Savkov <asavkov@redhat.com>
"funcs" in kpatch_create_patches_sections() and "entries" in
kpatch_create_kpatch_arch_section() were only used by sizeof, replaced
those with corresponding types.
Signed-off-by: Artem Savkov <asavkov@redhat.com>
Make sure symtab section was found before dereferencing it.
Found by covscan, see issue #984 for full log.
Signed-off-by: Artem Savkov <asavkov@redhat.com>
newdata variable is allocated through malloc call and requires a NULL
check.
Found by covscan, see issue #984 for full log.
Signed-off-by: Artem Savkov <asavkov@redhat.com>
Since ORC_STRUCT_SIZE is used for division in
kpatch_regenerate_orc_sections() we need to make sure that it is
properly set.
Found by covscan, see issue #984 for full log.
Signed-off-by: Artem Savkov <asavkov@redhat.com>
Make sure fixup section was found before dereferencing it.
Found by covscan, see issue #984 for full log.
Signed-off-by: Artem Savkov <asavkov@redhat.com>
Make sure rela_toc(1|2) are not null before dereferencing them in
rela_equal().
Found by covscan, see issue #984 for full log.
Signed-off-by: Artem Savkov <asavkov@redhat.com>
With the following patch:
diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index e008aefc3a9d..7c70e369390d 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -2228,6 +2228,8 @@ static void xs_tcp_shutdown(struct rpc_xprt *xprt)
struct socket *sock = transport->sock;
int skst = transport->inet ? transport->inet->sk_state : TCP_CLOSE;
+ asm("nop");
+
if (sock == NULL)
return;
switch (skst) {
We saw the following panic on a RHEL7.6 kernel:
Unable to handle kernel paging request for data at address 0xd00000000577f390
Faulting instruction address: 0xd000000002e918f4
Oops: Kernel access of bad area, sig: 11 [#1]
SMP NR_CPUS=2048 NUMA pSeries
Modules linked in: kpatch_3_10_0_957_1_3_1_1(OEK) nfsd nfs_acl rpcsec_gss_krb5 auth_rpcgss nfsv4 dns_resolver nfs lockd grace fscache sunrpc virtio_balloon ip_tables xfs libcrc32c virtio_net virtio_console virtio_blk virtio_pci virtio_ring virtio dm_mirror dm_region_hash dm_log dm_mod
CPU: 9 PID: 5961 Comm: kworker/9:1H Kdump: loaded Tainted: G OE K------------ 3.10.0-957.1.3.el7.ppc64le #1
Workqueue: xprtiod xprt_autoclose [sunrpc]
task: c00000000300c3c0 ti: c0000003f1814000 task.ti: c0000003f1814000
NIP: d000000002e918f4 LR: d000000002e57394 CTR: c00000000089d100
REGS: c0000003f1817980 TRAP: 0300 Tainted: G OE K------------ (3.10.0-957.1.3.el7.ppc64le)
MSR: 8000000100009033 <SF,EE,ME,IR,DR,RI,LE> CR: 240f2084 XER: 20000000
CFAR: 000000010bb5270c DAR: d00000000577f390 DSISR: 40000000 SOFTE: 1
GPR00: c00000000000b054 c0000003f1817c00 d00000000579add8 c000000214f0f4d0
GPR04: c0000003fd618200 c0000003fd618200 0000000000000001 0000000000000dc2
GPR08: 0000000000000dc3 0000000000000000 0000000000000000 d00000000577f370
GPR12: c0000003f1814000 c000000007b85100 c00000000012fd88 c0000003f711bb40
GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
GPR20: 0000000000000001 c0000000013510b0 0000000000000001 fffffffffffffef7
GPR24: 0000000000000000 0000000000000000 0000000000000000 c000000001b60600
GPR28: c000000214f0f000 c000000214f0f4d0 c000000214f0f408 c000000214f0f448
NIP [d000000002e918f4] __rpc_create_common.part.6+0x640/0x533c [sunrpc]
LR [d000000002e57394] xprt_autoclose+0x74/0xe0 [sunrpc]
Call Trace:
[c0000003f1817c00] [c00000000000b054] livepatch_handler+0x30/0x80 (unreliable)
[c0000003f1817c40] [c00000000012333c] process_one_work+0x1dc/0x680
[c0000003f1817ce0] [c000000000123980] worker_thread+0x1a0/0x520
[c0000003f1817d80] [c00000000012fe74] kthread+0xf4/0x100
[c0000003f1817e30] [c00000000000a628] ret_from_kernel_thread+0x5c/0xb4
Instruction dump:
396b4570 f8410018 e98b0020 7d8903a6 4e800420 00000000 73747562 000f49c0
c0000000 3d62fffe 396b4598 f8410018 <e98b0020> 7d8903a6 4e800420 00000000
---[ end trace 98e026b8fa880db7 ]---
The original version of xs_tcp_shutdown() has the following sequence:
0xd000000003cfda44 <xs_tcp_shutdown+148>: addi r1,r1,64
0xd000000003cfda48 <xs_tcp_shutdown+152>: ld r0,16(r1)
0xd000000003cfda4c <xs_tcp_shutdown+156>: ld r29,-24(r1)
0xd000000003cfda50 <xs_tcp_shutdown+160>: ld r30,-16(r1)
0xd000000003cfda54 <xs_tcp_shutdown+164>: ld r31,-8(r1)
0xd000000003cfda58 <xs_tcp_shutdown+168>: mtlr r0
0xd000000003cfda5c <xs_tcp_shutdown+172>: b 0xd000000003cfd768
That is, it restores the stack to the caller's stack frame and then does
a sibling call to the localentry point of xs_reset_transport()). So
when xs_reset_transport() returns, it will return straight to
xs_tcp_shutdown()'s caller (xprt_autoclose).
The patched version of the function has this instead (dumped from a
vmcore):
0xd000000003df0834 <xs_tcp_shutdown+148>: addi r1,r1,64
0xd000000003df0838 <xs_tcp_shutdown+152>: ld r0,16(r1)
0xd000000003df083c <xs_tcp_shutdown+156>: ld r29,-24(r1)
0xd000000003df0840 <xs_tcp_shutdown+160>: ld r30,-16(r1)
0xd000000003df0844 <xs_tcp_shutdown+164>: ld r31,-8(r1)
0xd000000003df0848 <xs_tcp_shutdown+168>: mtlr r0
0xd000000003df084c <xs_tcp_shutdown+172>: b 0xd000000003df0ad0
After restoring the stack, instead of branching directly to
xs_reset_transport(), it (rightfully) branches to a toc stub. A stub is
needed because the function it's branching to is in another module
(branching from the patch module to the sunrpc module).
The stub is:
0xd000000003df0ad0 <xs_tcp_shutdown+816>: addis r11,r2,-1
0xd000000003df0ad4 <xs_tcp_shutdown+820>: addi r11,r11,26328
0xd000000003df0ad8 <xs_tcp_shutdown+824>: std r2,24(r1)
0xd000000003df0adc <xs_tcp_shutdown+828>: ld r12,32(r11)
0xd000000003df0ae0 <xs_tcp_shutdown+832>: mtctr r12
0xd000000003df0ae4 <xs_tcp_shutdown+836>: bctr
And the "std r2,24(r1)" corrupts the caller's stack.
This stub makes sense for a normal call, because the stack would be
owned by the caller of the stub, so it's ok to write r2 to it. But
because this is a sibling call, the stack has been restored and r2 gets
incorrectly saved to the original caller's stack (i.e., xprt_autoclose's
stack).
So xprt_autoclose() -- which is in the sunrpc module -- gets the
livepatch module's toc pointer written to its stack. It panics on when
it tries to use that vlue on its very next call.
Fix it by disallowing sibling calls from patched functions on ppc64le.
In theory we could instead a) generate a custom stub, or b) modify the
kernel livepatch_handler code to save/restore the stack r2 value, but
this is easier for now.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Create an error if a patched function uses a jump label. We need this
until upstream livepatch supports them with a .klp.arch section.
Fixes#946.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
After Linux commit 47cdd64be483 ("dynamic_debug: refactor
dynamic_pr_debug and friends"), the name of the static local variable
used for dynamic printks is no longer "descriptor".
Make the is_special_static() check broader such that it doesn't care
about the variable name, and just whitelists any variable in the
__verbose section.
Fixes#990.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Starting with binutils 2.31, the Linux kernel may have R_X86_64_PLT32
relocations. Make sure we support them. This should be as simple as
treating R_X86_64_PLT32 exactly like R_X86_64_PC32 everywhere. For more
details see upstream commit torvalds/linux@b21ebf2.
This also fixes the following issue seen on Fedora 29:
```
$ kpatch-build/kpatch-build -t vmlinux ./test/integration/fedora-27/convert-global-local.patch
Using cache at /home/jpoimboe/.kpatch/src
Testing patch file(s)
Reading special section data
Building original source
Building patched source
Extracting new and modified ELF sections
ERROR: slub.o: 1 function(s) can not be patched
slub.o: function __kmalloc has no fentry/mcount call, unable to patch
/home/jpoimboe/git/kpatch/kpatch-build/create-diff-object: unreconcilable difference
ERROR: 1 error(s) encountered. Check /home/jpoimboe/.kpatch/build.log for more details.
```
Fixes#975.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Currently we do not support changes to functions referring to any of the
*_fixup sections on ppc64le. This patch introduces check for such
changes during the patchability check, where we abort building the
patch module.
This patch implements the phase 1 fix of 3 phases discussed at
https://github.com/dynup/kpatch/issues/974:
"
Phase 1 fix:
For kernel versions which don't have livepatch-specific powerpc code
(currently all kernel versions), kpatch-build needs to assert an error
if it detects that one of the following sections refers to a patched
function: __ftr_fixup, __mmu_ftr_fixup, __fw_ftr_fixup.
"
Suggested-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
Add support for __spec_barrier_fixup (barrier nospec fixup) special
section on ppc64le.
Suggested-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
Fix the size of special group __lwsync_fixup on ppc64le.
Suggested-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
.altinstr_replacement section may have relocation symbols which need to
be included, therefore we should call kpatch_include_symbol() to ensure
that its section is included as well.
The special section processing should also occur before
kpatch_print_changes() to provide accurate logging info.
Signed-off-by: Joe Lawrence <joe.lawrence@redhat.com>
The create-diff-object.c create intermediate ".kpatch.relocations"
sections instead of ".kpatch.dynrelas" sections, and add a new
section ".rela.kpatch.symbols", so we should update the conditions
in function kpatch_create_intermediate_sections for these changed.
Fixes: 87643703a7 ("create-diff-object: create .kpatch.relocations and .kpatch.symbols sections")
Signed-off-by: chenzefeng <chenzefeng2@huawei.com>
reason: Firstly, in the function lookup_open use the malloc to
allocate some memory, but call the function lookup_close
to free the memory.
Secondly, table->obj_sym->name, table->exp_sym->name and
table->exp_sym->objname used the strdup, so them should
free also.
Thirdly, adjust the order of make_nodname, if not, it
will cause an exception when free(exp_sym->objname) in
lookup_close.
Signed-off-by: chenzefeng <chenzefeng2@huawei.com>
This reverts commit 87c64519fc.
The jump label support doesn't work with upstream livepatch. Joe
Lawrence found the following ordering issue:
load_module
apply_relocations
/* Livepatch relocation sections are applied by livepatch */
if (info->sechdrs[i].sh_flags & SHF_RELA_LIVEPATCH)
continue;
post_relocation
module_finalize
jump_label_apply_nops << crash
...
do_init_module
do_one_initcall(mod->init)
__init patch_init [kpatch-patch]
klp_register_patch
klp_init_patch
klp_init_object
klp_init_object_loaded
klp_write_object_relocations
So jump_label_apply_nops() is called *before*
klp_write_object_relocations() has had a chance to write the klp
relocations (.klp.rela.kvm_intel.__jump_table, for example).
We need to resolve this upstream first.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Add support for jump labels, also known as static jumps, static keys,
static branches, and jump tables. Luckily,
kpatch_process_special_sections() is already generic enough to make this
an easy fix.
Fixes: #931
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
For fun I tried to create a livepatch of upstream patch
ad211f3e94b314a910d4af03178a0b52a7d1ee0a for my kernel. This
caused kpatch-build to fail with a NULL pointer derefence because
base_locals was NULL (returned via kpatch_elf_locals(), which
can return a NULL pointer). This patch fixes the SIGSEGV
via a NULL check. The end result is a live patch is created
and loaded.
Signed-off-by: Balbir singh <bsingharora@gmail.com>
kpatch_mark_ignored_sections include .rodata.str1.1 section but does
not include its section symbol, causing its section symbol can not be
included any more in kpatch_include_standard_elements. After the
section symbol is freed in kpatch_elf_teardown, we got a segmentation
fault in kpatch_create_intermediate_sections.
Signed-off-by: Zhipeng Xie <xiezhipeng1@huawei.com>
As discovered in #918, the `__FUNCTION__` static local variable is
similar to the `__func__` variable, in that it refers to the current
function name.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Sometimes due to config-dependency issues or other reasons whole
object-files would get optimized out from final vmlinux/module, in cases
like this create-diff-object would fail during symbol lookup table
creation in lookup_open(). Because lookup_open() call is situated before
we established that objectfile has changed this triggers not only on
real problems, but also during mass-rebulds caused by changes to
header-files. While it usually indicates a real issue with config this
should not prevent kpatch from building.
Move lookup_open() call so that it is called only for changed
object-files.
Fixes#910
Signed-off-by: Artem Savkov <asavkov@redhat.com>
strdup symbol names in kpatch_elf_locals and when noting down hint
instead of just copying pointers so that they are still usable after
we teardown/free kelf_base.
Signed-off-by: Artem Savkov <asavkov@redhat.com>
GCC puts the constant variable requiring relocation into .data.rel. or
.data.rel.ro depending upon the bind type of the symbol. Extend
is_bundledable() to check these .data sections too.
Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
4.18 adds -mcount-record to KBUILD_FLAGS when supported by the compiler.
This results in most of kpatch_create_mcount_sections()'s work being
already done, so we can at least skip the last part of it that updates
the first instruction in patched functions.
Signed-off-by: Artem Savkov <asavkov@redhat.com>
According to gcc8's man pages gcc can put functions into .text.unlikely
or .text.hot subfunctions during optimization. Add ".text.hot" to the
list of bundleable functions in is_bundleable().
Signed-off-by: Artem Savkov <asavkov@redhat.com>
gcc8 can place functions to .text.unlikely and .text.hot subsections
during optimizations. Allow symbols to change subsections instead of
failing.
Signed-off-by: Artem Savkov <asavkov@redhat.com>
Add a function that would detect parent/child symbol relations. So far
it only supports .cold.* symbols as children.
Signed-off-by: Artem Savkov <asavkov@redhat.com>
Building with GCC 7.3.0 on Debian sid fails with the following error:
gcc -g -O2 -fdebug-prefix-map=/build/kpatch-0.6.0=. -fstack-protector-strong -Wformat -Werror=format-security -MMD -MP -I../kmod/patch -Iinsn -Wall -Wsign-compare -g -Werror -Wdate-time -D_FORTIFY_SOURCE=2 -c -c
create-diff-object.c: In function 'kpatch_compare_correlated_rela_section':
create-diff-object.c:316:20: error: 'toc_data1' may be used uninitialized in this function [-Werror=maybe-uninitialized]
return toc_data1 == toc_data2;
~~~~~~~~~~^~~~~~~~~~~~
create-diff-object.c:256:16: note: 'toc_data1' was declared here
unsigned long toc_data1, toc_data2;
^~~~~~~~~
cc1: all warnings being treated as errors
This is a false positive as the code only compares those two values
after initializing them. But lets keep GCC happy.
Signed-off-by: Simon Ruderich <simon@ruderich.org>
Convert magic exit status values into a common enum for clarity.
Suggested-by: Artem Savkov <asavkov@redhat.com>
Signed-off-by: Joe Lawrence <joe.lawrence@redhat.com>
Change the "FILE symbol not found in base. Stripped?" (fatal) error
message into a warning. These crop up whenever a change is made to an
assembly file.
Signed-off-by: Joe Lawrence <joe.lawrence@redhat.com>
Finally add support for processing the ORC unwinder sections.
The ORC unwinder sections are more special than the other special
sections, so they need their own dedicated function to process them,
though the code is similar to kpatch_regenerate_special_sections().
BTW, upstream livepatch still doesn't support the ORC unwinder. That
change will be coming soon (probably Linux 4.19).
Fixes#785.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Since the codeset supports just the 64 bit variant, lets move
to __powerpc64__ and use it. I checked the ABI doc as well
and the kernel/gcc.
Signed-off-by: Balbir singh <bsingharora@gmail.com>
gcc8 introduces ".cold." optimization symbols that have arbitrary
trainling numbers in their names just like ".isra." and others.
Add ".cold." to a condition in kpatch_rename_mangled_functions()
Signed-off-by: Artem Savkov <asavkov@redhat.com>
When I made a patch to the nfsd module on a ppc64le system with a RHEL 7
based kernel, livepatch prevented the target module from loading with:
livepatch: symbol '.TOC.' not found in symbol table
References to this symbol are treated specially by the kernel module
loader, so references to it should never be converted to dynrelas.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
While building a gcc-consprop patch from integration tests gcc8 would place a
__timekeeping_inject_sleeptime.constprop.18.cold.27 symbol into
.text.unlikely.__timekeeping_inject_sleeptime.constprop.18 section. Because
section name doesn't have the '.cold.27' suffix this symbol fails
is_bundleable() check while still being bundleable and later exits early in
kpatch_rename_mangled_functions() without renaming the corresponding patched
function. All of this results in a create-diff-object errror:
ERROR: timekeeping.o: symbol changed sections: __timekeeping_inject_sleeptime.constprop.18.cold.27
/home/asavkov/dev/kpatch/kpatch-build/create-diff-object: unreconcilable difference
Fix by ignoring .cold.* name suffix in is_bundleable() for.text.unlikely
sections.
Signed-off-by: Artem Savkov <asavkov@redhat.com>
create-diff-object doesn't really need the full kernel object file as
input. All it requires is a symbol table. Switch to using "eu-readelf -s"'s
output instead of object files. This will enable us to cover more cases
in unit tests.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Artem Savkov <asavkov@redhat.com>
Use named variables instead of cmdline argument indexes so that it is
easier to read create-diff-object.c.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Artem Savkov <asavkov@redhat.com>
ppc64le lacked kpatch_line_macro_change_only() implementation. Add one
based on x86_64 version and available unit tests.
Fixes: #843
Signed-off-by: Artem Savkov <asavkov@redhat.com>
Update the `__LINE__` macro detection heuristics based on recent kernel
changes:
- __warn_printk(), used in the WARN macros, which put the line number in
%edx
- __might_fault(), used in the usercopy macros
- lockdep_rcu_suspicious(), used in the RCU_LOCKDEP_WARN macro
- printk(), used in the ASSERT_RTNL macro
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
With #755, we started using dynrelas for function pointers. However,
this behavior only makes sense for function pointers to existing
functions. For function pointers to *new* functions, just use a normal
rela.
The 'function-ptr-new' unit test is from the following patch:
https://github.com/dynup/kpatch/files/1927198/new-static-callback.patch.txtFixes#834.
Fixes: 495e619750 ("kpatch-build, x86: do not use the patched functions as callbacks directly")
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
- convert section/symbol indexes and rela->offset to unsigned int as I
couldn't find any way for them to become negative.
- cast a number of rela->addend comparisons to int (assuming an 64bit
system this should be enough)
- a number of simple for-loop counter conversions to the type it
compares against
Signed-off-by: Artem Savkov <asavkov@redhat.com>
Because of signedness difference kpatch_check_relocations() would trigger
an error on any negative addend.
Fix by casting Elf_Data->d_size from size_t to int.
Signed-off-by: Artem Savkov <asavkov@redhat.com>
Check that none of the relocations are out-of-range of their
corresponding sections before writing the output elf.
Fixes: #618
Signed-off-by: Artem Savkov <asavkov@redhat.com>
Heavy lifting of reading .toc rela entries for rela symbols
referring to .toc + offset, can be simplified using toc_rela() in
rela_equal() and remove the #ifdery guarding PowerPC code.
This patch also trims the commentary related to PowerPC.
Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
On ppc64le, the static local variable correlation doesn't take into
account the .toc rela indirection for data references, meaning that it's
basically broken in many cases.
Fix it by making the code .toc-aware.
Fixes#793.
Reported-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Upstream 4.15 kernels provide support for pre and post (un)patch
callbacks, inspired by the kpatch load hooks. Add support for them
in the livepatch-patch-hook.
At the same time, convert the kpatch hooks to use the same API.
Signed-off-by: Joe Lawrence <joe.lawrence@redhat.com>
kpatch_include_symbol() is confusing. Refactor it:
- Remove the "inclusion tree" debug messages. I never use them, and
they just help make the code more confusing and the debug output more
cluttered.
- Split it up into two functions: kpatch_include_symbol() and
kpatch_include_section(), so that kpatch_include_section() can be used
elsewhere.
- Call kpatch_include_section() from kpatch_include_standard_elements().
This covertly fixes#702, by also including the .rela.rodata section.
- Add a bunch of comments to clarify some of the trickier points.
Fixes#702.
Fixes#807.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
It was observed by Evgenii Shatokhin in PR#755, that when the RCU
callback was called on the patched function, from unloaded livepatch
module triggered a kernel crash.
This patch implements the approach on PowerPC outlined in PR#755.
With -mcmodel=large, like any other data, function pointers are also
loaded relative to the current TOC base and are populated as
relocation entries in .toc section. Every function passing a function
pointer as the argument need to load the function address through
.toc section + offset. Convert such .toc + offset relocation into
a dynamic rela, which resolves to original function address, during
module load.
Also move the comment related to nested function check, into
may_need_dynrela().
Suggested-by: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Evgenii Shatokhin <eshatokhin@virtuozzo.com>
Cc: Joe Lawrence <jdl1291@gmail.com>
Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
A kernel crash happened in __do_softirq() in very rare cases when the
binary patch created from mainline commit be82485fbcbb
("netlink: fix an use-after-free issue for nlk groups") was unloaded.
Investigation has shown that the kernel tried to execute an RCU
callback, deferred_put_nlk_sk(), defined in the patch module after the
module had been unloaded.
The callback was set by the patched variant of netlink_release() and
the address of the patched deferred_put_nlk_sk() was used, rather than
the address of the original function.
Similar problems occur with workqueue functions as well.
As suggested in
https://github.com/dynup/kpatch/pull/755#issuecomment-344135224,
create-diff-object was modified so that the addresses of the original
functions were used in such situations, at least for x86 systems. A
similar fix for PowerPC was added as well.
Changes in v4:
* '#ifdef __x86_64__' was removed. It is not actually needed right now
because the constants for relocation types are different on different
architectures.
Changes in v3:
* Minor refactoring and a comment explaining what this all is about.
Quite lengthy, but the dynrela-related code is really far from obvious.
Changes in v2:
* Handle the nested functions the same way as before, because they are
unlikely to be used as asynchronous callbacks. Example: cmp() in
bch_cache_show() from drivers/md/bcache/sysfs.c in the kernel 4.4.
As the nested functions are local to the functions they are defined in,
the compiler names them in a similar way to static locals: <name>.<number>.
Currently, we filter out all functions with '.' in their names. If there
are any asynchronous callbacks in the kernel that have a dot in their
names too, they could be handled in the future patches. It is unclear
though, if the callbacks with such names can appear in the kernel.
Signed-off-by: Evgenii Shatokhin <eshatokhin@virtuozzo.com>
With kernel commit b1fca27d384 ("kernel debug: support resetting
WARN*_ONCE") the *_ONCE warnings are placed .data.once section.
Including .data.once section is valid, so add an check in
kpatch_verify_patchability() while checking for included invalid
sections.
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
Symbols with R_PPC64_REL24 relocation type are functions and it's
currently assumed that all functions are replaced with their respective
section symbols.
There are function whose reference are not straight forward section
symbol but section + offset. These function replacement should be
handled more like bundled sections. Remove the check, which imposes
the inital assumption.
Suggested-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
.toc section entries are mostly place holder for relocation entries,
specified in .rela.toc section. Sometimes, .toc section may have
constants as entries. These constants are not reference to any symbols,
but plain instructions mostly due to some arthimetics in the functions
referring them.
They are referred by the functions like normal .toc entries, these
entries can not be resolved to any symbols. This patch creates a list
of constants if available for .toc sections and compares them in
rela_equal() to ensure their is no mismatch in the generated constants
for original and patched .o files.
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
GCC 7.2.1 complains about the usage of the PPC64_LOCAL_ENTRY_OFFSET
macro:
create-diff-object.c: In function ‘is_gcc6_localentry_bundled_sym’:
create-diff-object.c:119:83: error: ‘<<’ in boolean context, did you mean ‘<’ ? [-Werror=int-in-bool-context]
(((1 << (((other) & STO_PPC64_LOCAL_MASK) >> STO_PPC64_LOCAL_BIT)) >> 2) << 2)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~
create-diff-object.c:140:10: note: in expansion of macro ‘PPC64_LOCAL_ENTRY_OFFSET’
return (PPC64_LOCAL_ENTRY_OFFSET(sym->sym.st_other) &&
^~~~~~~~~~~~~~~~~~~~~~~~
Fix it by explicitly treating the macro as an integer instead of a bool.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Found in the scope of https://github.com/dynup/kpatch/pull/755 but not
related to the main problem discussed there.
kpatch_create_patches_sections() and kpatch_create_intermediate_sections()
used 'hint' in error messages.
However, the string 'hint' refers to is owned by 'kelf_base' and is
freed before kpatch_create_*_sections() are called. As a result, if
these functions try to output errors and print 'hint',
create-diff-object will crash.
As suggested in the mentioned PR, 'hint' is actually no longer needed at
that stage, so I have removed it from kpatch_create_*_sections().
When creating .kpatch.relocations, there's no reason to convert the
relocation destinations to symbols. In fact, it's actively harmful
because it makes it harder for create-klp-module to deal with the GCC 6+
8-byte localentry gap.
This also fixes a regression which was introduced in 5888f316e6, which
broke ppc64le relocations.
Fixes#754.
Fixes: 5888f316e6 ("create-klp-module: support unbundled symbols")
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
kpatch_relocation's 'dest' addend and 'offset' fields are redundant. In
fact, the 'offset' field isn't always accurate because it doesn't have a
relocation, so its value doesn't adjust when multiple .o files are
combined. Just use the 'dest' addend instead.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
kpatch_replace_sections_syms() assumes that all bundled symbols start at
section offset zero. With ppc64le and GCC 6+, that assumption is no
longer accurate. When replacing a rela symbol section with its
corresponding symbol, adjust the addend as necessary.
Also, with this fix in place, the workaround in
create_klp_relasecs_and_syms() can be removed.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
On ppc64le, adding a printk to total_mapping_size() caused it to change
from non-localentry to localentry, presumably because it was no longer a
leaf function. With GCC 6, a localentry function is offset by 8 in the
section, so different st_values are ok.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
The STT_FUNC and SHN_UNDEF checks aren't needed because they're already
implied by the localentry check.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
is_localentry_sym() isn't quite the right name, because it also checks
for the 8-byte gap introduced by GCC 6, and also checks that the
function is otherwise at the beginning of the section.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
The paravirt_patch_site struct has 12 bytes of data and 4 bytes of
padding, for a total of 16 bytes. However, when laying out the structs
in the .parainstructions section, the vmlinux script only aligns before
each struct's data, not after. So the last entry doesn't have the
4-byte padding, which breaks kpatch_regenerate_special_section()'s
assumption of a 16-byte struct, resulting in a memcpy past the end of
the section.
Fixes#747.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
