| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <f28b0c22-af83-31bf-8416-e896783fc105@gmx.de>
Date: Sat, 29 Jan 2022 18:50:10 +0100
From: Helge Deller <deller@....de>
To: kernel test robot <lkp@...el.com>
Cc: kbuild-all@...ts.01.org, linux-kernel@...r.kernel.org,
John David Anglin <dave.anglin@...l.net>
Subject: Re: [deller-parisc:5.17-vdso-mini-3 1/1]
arch/parisc/kernel/vdso32/sigtramp.S:39: Error: unknown pseudo-op: `.proc'
On 1/29/22 03:00, kernel test robot wrote:
> tree: https://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux.git 5.17-vdso-mini-3
> head: 8d7e3770fb1fb6c01147f8ebf2914cd2037fac16
> commit: 8d7e3770fb1fb6c01147f8ebf2914cd2037fac16 [1/1] parisc: Add vDSO support
> config: parisc-generic-64bit_defconfig (https://download.01.org/0day-ci/archive/20220129/202201290925.4WPmV6eS-lkp@intel.com/config)
> compiler: hppa64-linux-gcc (GCC) 11.2.0
> reproduce (this is a W=1 build):
> wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
> chmod +x ~/bin/make.cross
> # https://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux.git/commit/?id=8d7e3770fb1fb6c01147f8ebf2914cd2037fac16
> git remote add deller-parisc https://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux.git
> git fetch --no-tags deller-parisc 5.17-vdso-mini-3
> git checkout 8d7e3770fb1fb6c01147f8ebf2914cd2037fac16
> # save the config file to linux build tree
> mkdir build_dir
> COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-11.2.0 make.cross O=build_dir ARCH=parisc prepare
I'm somewhat confused by those errors.
There are two changes which happened in the parisc port:
1.) If you want to build a 64bit kernel, you now need "ARCH=parisc64". Before "ARCH=parisc" was sufficient,
but now ARCH=parisc64 builds 64bit kernel, and ARCH=parisc builds the 32bit kernel.
2.) The patch which is being complained about here is, that it adds an initial vDSO, which
includes builds the 64- and 32-bit vDSO in the same build.
So, on a 64bit kernel build (which seems to happen here), the 32-bit gcc-11 compiler needs to be
installed as well in order to be able to build the 32-bit "compat" vDSO.
Maybe the errors here are happening, because the 32bit hppa cross compiler is missing?
Could you please check ?
Helge
> All errors (new ones prefixed by >>):
>
> arch/parisc/kernel/vdso64/Makefile:30: FORCE prerequisite is missing
> arch/parisc/kernel/vdso32/sigtramp.S: Assembler messages:
>>> arch/parisc/kernel/vdso32/sigtramp.S:39: Error: unknown pseudo-op: `.proc'
>>> arch/parisc/kernel/vdso32/sigtramp.S:40: Error: unknown pseudo-op: `.callinfo'
>>> arch/parisc/kernel/vdso32/sigtramp.S:41: Error: unknown pseudo-op: `.entry'
>>> arch/parisc/kernel/vdso32/sigtramp.S:44: Error: no such instruction: `ldi 0,%r25'
>>> arch/parisc/kernel/vdso32/sigtramp.S:45: Error: no such instruction: `ldi 173,%r20'
>>> arch/parisc/kernel/vdso32/sigtramp.S:46: Error: no such instruction: `ble 0x100(%sr2,%r0)'
> arch/parisc/kernel/vdso32/sigtramp.S:49: Error: no such instruction: `ldi 1,%r25'
> arch/parisc/kernel/vdso32/sigtramp.S:50: Error: no such instruction: `ldi 173,%r20'
> arch/parisc/kernel/vdso32/sigtramp.S:51: Error: no such instruction: `ble 0x100(%sr2,%r0)'
>>> arch/parisc/kernel/vdso32/sigtramp.S:54: Error: unknown pseudo-op: `.exit'
>>> arch/parisc/kernel/vdso32/sigtramp.S:55: Error: unknown pseudo-op: `.procend'
>>> arch/parisc/kernel/vdso32/sigtramp.S:76: Error: unknown pseudo-op: `.stringz'
> make[2]: *** [arch/parisc/kernel/vdso32/Makefile:34: arch/parisc/kernel/vdso32/sigtramp.o] Error 1
> arch/parisc/kernel/vdso32/restart_syscall.S: Assembler messages:
>>> arch/parisc/kernel/vdso32/restart_syscall.S:16: Error: bad or irreducible absolute expression
>>> arch/parisc/kernel/vdso32/restart_syscall.S:16: Error: junk at end of line, first unrecognized character is `:'
>>> arch/parisc/kernel/vdso32/restart_syscall.S:26: Error: no such instruction: `ldw 0(%sp),%r31'
>>> arch/parisc/kernel/vdso32/restart_syscall.S:29: Error: no such instruction: `be 0x100(%sr2,%r0)'
>>> arch/parisc/kernel/vdso32/restart_syscall.S:30: Error: no such instruction: `ldi 0,%r20'
>>> arch/parisc/kernel/vdso32/restart_syscall.S:32: Error: .cfi_endproc without corresponding .cfi_startproc
> make[2]: *** [arch/parisc/kernel/vdso32/Makefile:34: arch/parisc/kernel/vdso32/restart_syscall.o] Error 1
> make[2]: Target 'include/generated/vdso32-offsets.h' not remade because of errors.
> make[1]: *** [arch/parisc/Makefile:186: vdso_prepare] Error 2
> make[1]: Target 'prepare' not remade because of errors.
> make: *** [Makefile:219: __sub-make] Error 2
> make: Target 'prepare' not remade because of errors.
>
>
> vim +39 arch/parisc/kernel/vdso32/sigtramp.S
>
> 12
> 13 .text
> 14
> 15 /* Gdb expects the trampoline is on the stack and the pc is offset from
> 16 a 64-byte boundary by 0, 4 or 5 instructions. Since the vdso trampoline
> 17 is not on the stack, we need a new variant with different offsets and
> 18 data to tell gdb where to find the signal context on the stack.
> 19
> 20 Here we put the offset to the context data at the start of the trampoline
> 21 region and offset the first trampoline by 2 instructions. Please do
> 22 not change the trampoline as the code in gdb depends on the following
> 23 instruction sequence exactly.
> 24 */
> 25 .align 64
> 26 .word SIGFRAME_CONTEXT_REGS32
> 27
> 28 /* The nop here is a hack. The dwarf2 unwind routines subtract 1 from
> 29 the return address to get an address in the middle of the presumed
> 30 call instruction. Since we don't have a call here, we artifically
> 31 extend the range covered by the unwind info by adding a nop before
> 32 the real start.
> 33 */
> 34 nop
> 35
> 36 .globl __kernel_sigtramp_rt
> 37 .type __kernel_sigtramp_rt, @function
> 38 __kernel_sigtramp_rt:
> > 39 .proc
> > 40 .callinfo FRAME=ASM_SIGFRAME_SIZE32,CALLS,SAVE_RP
> > 41 .entry
> 42
> 43 .Lsigrt_start = . - 4
> > 44 0: ldi 0, %r25 /* (in_syscall=0) */
> > 45 ldi __NR_rt_sigreturn, %r20
> > 46 ble 0x100(%sr2, %r0)
> 47 nop
> 48
> 49 1: ldi 1, %r25 /* (in_syscall=1) */
> 50 ldi __NR_rt_sigreturn, %r20
> 51 ble 0x100(%sr2, %r0)
> 52 nop
> 53 .Lsigrt_end:
> > 54 .exit
> > 55 .procend
> 56 .size __kernel_sigtramp_rt,.-__kernel_sigtramp_rt
> 57
> 58
> 59 .section .eh_frame,"a",@progbits
> 60
> 61 /* This is where the mcontext_t struct can be found on the stack. */
> 62 #define PTREGS SIGFRAME_CONTEXT_REGS32 /* 32-bit process offset is -672 */
> 63
> 64 /* Register REGNO can be found at offset OFS of the mcontext_t structure. */
> 65 .macro rsave regno,ofs
> 66 .byte 0x05 /* DW_CFA_offset_extended */
> 67 .uleb128 \regno; /* regno */
> 68 .uleb128 \ofs /* factored offset */
> 69 .endm
> 70
> 71 .Lcie:
> 72 .long .Lcie_end - .Lcie_start
> 73 .Lcie_start:
> 74 .long 0 /* CIE ID */
> 75 .byte 1 /* Version number */
> > 76 .stringz "zRS" /* NUL-terminated augmentation string */
Powered by blists - more mailing lists