user/sven/linux.git/scripts/mod/modpost.c, branch v4.2.4

modpost: work correctly with tile coldtext sections

2015-07-08T22:53:49Z

The tilegx and tilepro compilers use .coldtext for their unlikely executed text section name, so an __attribute__((cold)) function will (when compiled with higher optimization levels) land in the .coldtext section. Modify modpost to add .coldtext to the set of OTHER_TEXT_SECTIONS so we don't get warnings about referencing such a section in an __ex_table block, and then also modify arch/tile/lib/memcpy_user_64.c so that it uses plain ".coldtext" instead of ".coldtext.memcpy". The latter naming is a relic of an earlier use of -ffunction-sections, which we no longer use by default. Signed-off-by: Chris Metcalf Acked-by: Rusty Russell

modpost: don't emit section mismatch warnings for compiler optimizations

2015-04-22T08:01:34Z

Currently an allyesconfig build [gcc-4.9.1] can generate the following: WARNING: vmlinux.o(.text.unlikely+0x3864): Section mismatch in reference from the function cpumask_empty.constprop.3() to the variable .init.data:nmi_ipi_mask which comes from the cpumask_empty usage in arch/x86/kernel/nmi_selftest.c. Normally we would not see a symbol entry for cpumask_empty since it is: static inline bool cpumask_empty(const struct cpumask *srcp) however in this case, the variant of the symbol gets emitted when GCC does constant propagation optimization. Fix things up so that any locally optimized constprop variants don't warn when accessing variables that live in the __init sections. Signed-off-by: Paul Gortmaker Signed-off-by: Rusty Russell

modpost: expand pattern matching to support substring matches

2015-04-22T08:01:33Z

Currently the match() function supports a leading * to match any prefix and a trailing * to match any suffix. However there currently is not a combination of both that can be used to target matches of whole families of functions that share a common substring. Here we expand the *foo and foo* match to also support *foo* with the goal of targeting compiler generated symbol names that contain strings like ".constprop." and ".isra." Signed-off-by: Paul Gortmaker Signed-off-by: Rusty Russell

modpost: do not try to match the SHT_NUL section.

2015-04-22T08:01:33Z

Trying to match the SHT_NUL section isn't useful and causes build failures on parisc and mn10300 since the addition of section strict white-listing and __ex_table sanitizing. Signed-off-by: Quentin Casasnovas Reported-by: Guenter Roeck Fixes: 050e57fd5936 ("modpost: add strict white-listing when referencing....") Fixes: 52dc0595d540 ("modpost: handle relocations mismatch in __ex_table.") Tested-by: Guenter Roeck Signed-off-by: Rusty Russell

modpost: fix extable entry size calculation.

2015-04-22T08:01:32Z

As Guenter pointed out, we were never really calculating the extable entry size because the pointer arithmetic was simply wrong. We want to check we're handling the second relocation in __ex_table to infer an entry size, but we were using (void*) pointers instead of Elf_Rel[a]* ones. This fixes the problem by moving that check in the caller (since we can deal with different types of relocations) and add is_second_extable_reloc() to make the whole thing more readable. Signed-off-by: Quentin Casasnovas Reported-by: Guenter Roeck CC: Rusty Russell Signed-off-by: Rusty Russell

modpost: fix inverted logic in is_extable_fault_address().

2015-04-22T08:01:31Z

As Guenter pointed out, we want to assert that extable_entry_size has been discovered and not the other way around. Moreover, this sanity check is only valid when we're not dealing with the first relocation in __ex_table, since we have not discovered the extable entry size at that point. This was leading to a divide-by-zero on some architectures and make the build fail. Signed-off-by: Quentin Casasnovas Reported-by: Guenter Roeck CC: Rusty Russell Signed-off-by: Rusty Russell

modpost: handle -ffunction-sections

2015-04-22T08:01:31Z

52dc0595d540 introduced OTHER_TEXT_SECTIONS for identifying what sections could validly have __ex_table entries. Unfortunately, it wasn't tested with -ffunction-sections, which some architectures use. Reported-by: kbuild test robot Cc: Quentin Casasnovas Signed-off-by: Rusty Russell

modpost: Whitelist .text.fixup and .exception.text

2015-04-22T08:01:20Z

32-bit and 64-bit ARM use these sections to store executable code, so they must be whitelisted in modpost's table of valid text sections. Signed-off-by: Thierry Reding Signed-off-by: Rusty Russell

modpost: document the use of struct section_check.

2015-04-13T11:33:04Z

struct section_check is used as a generic way of describing what relocations are authorized/forbidden when running modpost. This commit tries to describe how each field is used. Signed-off-by: Quentin Casasnovas Signed-off-by: Rusty Russell (Fixed "mist"ake)

modpost: handle relocations mismatch in __ex_table.

2015-04-13T11:33:03Z

__ex_table is a simple table section where each entry is a pair of addresses - the first address is an address which can fault in kernel space, and the second address points to where the kernel should jump to when handling that fault. This is how copy_from_user() does not crash the kernel if userspace gives a borked pointer for example. If one of these addresses point to a non-executable section, something is seriously wrong since it either means the kernel will never fault from there or it will not be able to jump to there. As both cases are serious enough, we simply error out in these cases so the build fails and the developper has to fix the issue. In case the section is executable, but it isn't referenced in our list of authorized sections to point to from __ex_table, we just dump a warning giving more information about it. We do this in case the new section is executable but isn't supposed to be executed by the kernel. This happened with .altinstr_replacement, which is executable but is only used to copy instructions from - we should never have our instruction pointer pointing in .altinstr_replacement. Admitedly, a proper fix in that case would be to just set .altinstr_replacement NX, but we need to warn about future cases like this. Signed-off-by: Quentin Casasnovas Signed-off-by: Rusty Russell (added long casts)