lkml.org 
[lkml]   [2012]   [Feb]   [22]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    SubjectRe: [PATCH 1/3] uprobes/core: Make instruction tables volatile.
    > * Srikar Dronamraju <srikar@linux.vnet.ibm.com> wrote:
    >
    > > From: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
    > >
    > > Some versions of gcc spits a warning about the asm operand for test_bit and
    > > also causes the first long of the instruction table to be output.
    > >
    > > Fix is similar to 7115e3fc on arch/x86/kernel/kprobes.c
    > >
    > > Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
    > > ---
    > > arch/x86/kernel/uprobes.c | 61 ++++++++++++++++++++++++---------------------
    > > 1 files changed, 32 insertions(+), 29 deletions(-)
    > >
    > > diff --git a/arch/x86/kernel/uprobes.c b/arch/x86/kernel/uprobes.c
    > > index cf2a184..13d616d 100644
    > > --- a/arch/x86/kernel/uprobes.c
    > > +++ b/arch/x86/kernel/uprobes.c
    > > @@ -53,34 +53,12 @@
    > > (bc##UL << 0xc)|(bd##UL << 0xd)|(be##UL << 0xe)|(bf##UL << 0xf)) \
    > > << (row % 32))
    > >
    > > -#ifdef CONFIG_X86_64
    > > -static u32 good_insns_64[256 / 32] = {
    > > - /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
    > > - /* ---------------------------------------------- */
    > > - W(0x00, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0) | /* 00 */
    > > - W(0x10, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0) , /* 10 */
    > > - W(0x20, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0) | /* 20 */
    > > - W(0x30, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0) , /* 30 */
    > > - W(0x40, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) | /* 40 */
    > > - W(0x50, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* 50 */
    > > - W(0x60, 0, 0, 0, 1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0) | /* 60 */
    > > - W(0x70, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* 70 */
    > > - W(0x80, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) | /* 80 */
    > > - W(0x90, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* 90 */
    > > - W(0xa0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) | /* a0 */
    > > - W(0xb0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* b0 */
    > > - W(0xc0, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0) | /* c0 */
    > > - W(0xd0, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* d0 */
    > > - W(0xe0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0) | /* e0 */
    > > - W(0xf0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1) /* f0 */
    > > - /* ---------------------------------------------- */
    > > - /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
    > > -};
    > > -#endif
    > > -
    > > -/* Good-instruction tables for 32-bit apps */
    > > -
    > > -static u32 good_insns_32[256 / 32] = {
    > > +/*
    > > + * Good-instruction tables for 32-bit apps. This is non-const and volatile
    > > + * to keep gcc from statically optimizing it out, as variable_test_bit makes
    > > + * some versions of gcc to think only *(unsigned long*) is used.
    > > + */
    > > +static volatile u32 good_insns_32[256 / 32] = {
    > > /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
    > > /* ---------------------------------------------- */
    > > W(0x00, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 0) | /* 00 */
    > > @@ -104,7 +82,7 @@ static u32 good_insns_32[256 / 32] = {
    > > };
    > >
    > > /* Using this for both 64-bit and 32-bit apps */
    > > -static u32 good_2byte_insns[256 / 32] = {
    > > +static volatile u32 good_2byte_insns[256 / 32] = {
    > > /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
    > > /* ---------------------------------------------- */
    > > W(0x00, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1) | /* 00 */
    > > @@ -127,6 +105,31 @@ static u32 good_2byte_insns[256 / 32] = {
    > > /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
    > > };
    > >
    > > +#ifdef CONFIG_X86_64
    > > +/* Good-instruction tables for 64-bit apps */
    > > +static volatile u32 good_insns_64[256 / 32] = {
    > > + /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
    > > + /* ---------------------------------------------- */
    > > + W(0x00, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0) | /* 00 */
    > > + W(0x10, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0) , /* 10 */
    > > + W(0x20, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0) | /* 20 */
    > > + W(0x30, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0) , /* 30 */
    > > + W(0x40, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) | /* 40 */
    > > + W(0x50, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* 50 */
    > > + W(0x60, 0, 0, 0, 1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0) | /* 60 */
    > > + W(0x70, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* 70 */
    > > + W(0x80, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) | /* 80 */
    > > + W(0x90, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* 90 */
    > > + W(0xa0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) | /* a0 */
    > > + W(0xb0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* b0 */
    > > + W(0xc0, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0) | /* c0 */
    > > + W(0xd0, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* d0 */
    > > + W(0xe0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0) | /* e0 */
    > > + W(0xf0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1) /* f0 */
    > > + /* ---------------------------------------------- */
    > > + /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
    > > +};
    > > +#endif
    >
    > Hm, why was the table moved around? This makes it hard to see
    > whether it's just a single volatile that is added, or something
    > more.
    >

    Without the table moving, the volatile was first used under #ifdef
    CONFIG_X86_64. I was wondering if I add a comment in #ifdef people
    looking at it might assume that its a problem only on x86_64 only.

    So I moved the CONFIG_X86_64 specific table down and added the comment
    for volatile at the first instance its used.

    --
    Thanks and Regards
    Srikar



    \
     
     \ /
      Last update: 2012-02-22 11:07    [W:0.027 / U:33.556 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site