| Message ID | 1418020212-4303-1-git-send-email-khandual@linux.vnet.ibm.com (mailing list archive) |
|---|---|
| State | Changes Requested |
| Delegated to: | Michael Ellerman |
| Headers | show |
On Mon, 2014-08-12 at 06:30:08 UTC, Anshuman Khandual wrote: > Currently DSCR (Data Stream Control Register) can be accessed with > mfspr or mtspr instructions inside a thread via two different SPR > numbers. One being the user accessible problem state SPR number 0x03 > and the other being the privilege state SPR number 0x11. All access > through the privilege state SPR number get emulated through illegal > instruction exception. Any access through the problem state SPR number > raises one facility unavailable exception which sets the thread based > dscr_inherit bit and enables DSCR facility through FSCR register thus > allowing direct access to DSCR without going through this exception in > the future. We set the thread.dscr_inherit bit whether the access was > with mfspr or mtspr instruction which is neither correct nor does it > match the behaviour through the instruction emulation code path driven > from privilege state SPR number. User currently observes two different > kind of behaviour when accessing the DSCR through these two SPR numbers. > This problem can be observed through these two test cases by replacing > the privilege state SPR number with the problem state SPR number. > > (1) http://ozlabs.org/~anton/junkcode/dscr_default_test.c > (2) http://ozlabs.org/~anton/junkcode/dscr_explicit_test.c Can you convert those into a selftest please? cheers
On 12/09/2014 03:41 PM, Michael Ellerman wrote: > On Mon, 2014-08-12 at 06:30:08 UTC, Anshuman Khandual wrote: >> Currently DSCR (Data Stream Control Register) can be accessed with >> mfspr or mtspr instructions inside a thread via two different SPR >> numbers. One being the user accessible problem state SPR number 0x03 >> and the other being the privilege state SPR number 0x11. All access >> through the privilege state SPR number get emulated through illegal >> instruction exception. Any access through the problem state SPR number >> raises one facility unavailable exception which sets the thread based >> dscr_inherit bit and enables DSCR facility through FSCR register thus >> allowing direct access to DSCR without going through this exception in >> the future. We set the thread.dscr_inherit bit whether the access was >> with mfspr or mtspr instruction which is neither correct nor does it >> match the behaviour through the instruction emulation code path driven >> from privilege state SPR number. User currently observes two different >> kind of behaviour when accessing the DSCR through these two SPR numbers. >> This problem can be observed through these two test cases by replacing >> the privilege state SPR number with the problem state SPR number. >> >> (1) http://ozlabs.org/~anton/junkcode/dscr_default_test.c >> (2) http://ozlabs.org/~anton/junkcode/dscr_explicit_test.c > > Can you convert those into a selftest please? Hey Michael, Yeah I wanted to convert all these tests which are related to DSCR into individual self tests for powerpc. All these test cases have Anton Blanchard and IBM's copyright on it but they are licensed with GPL V2. Not sure whether Anton needs to okay this before I can modify them for self tests, put his (Anton's) signed-off-by and then post it in the mailing list. (1) http://ozlabs.org/~anton/junkcode/dscr_default_test.c (2) http://ozlabs.org/~anton/junkcode/dscr_explicit_test.c (3) http://ozlabs.org/~anton/junkcode/dscr_inherit_exec_test.c (4) http://ozlabs.org/~anton/junkcode/dscr_inherit_test.c (5) http://ozlabs.org/~anton/junkcode/user_dscr_test.c
Hi Anshuman, > Yeah I wanted to convert all these tests which are related to DSCR > into individual self tests for powerpc. All these test cases have > Anton Blanchard and IBM's copyright on it but they are licensed with > GPL V2. Not sure whether Anton needs to okay this before I can modify > them for self tests, put his (Anton's) signed-off-by and then post it > in the mailing list. > > (1) http://ozlabs.org/~anton/junkcode/dscr_default_test.c > (2) http://ozlabs.org/~anton/junkcode/dscr_explicit_test.c > (3) http://ozlabs.org/~anton/junkcode/dscr_inherit_exec_test.c > (4) http://ozlabs.org/~anton/junkcode/dscr_inherit_test.c > (5) http://ozlabs.org/~anton/junkcode/user_dscr_test.c I'm happy for the testcases to be modified and merged into the kernel! When you submit them I'll give them a once over and attach my Signed-off-by. Thanks, Anton
diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c index 0dc43f9..d85a3e7 100644 --- a/arch/powerpc/kernel/traps.c +++ b/arch/powerpc/kernel/traps.c @@ -1377,6 +1377,7 @@ void facility_unavailable_exception(struct pt_regs *regs) }; char *facility = "unknown"; u64 value; + u32 instword; u8 status; bool hv; @@ -1388,12 +1389,44 @@ void facility_unavailable_exception(struct pt_regs *regs) status = value >> 56; if (status == FSCR_DSCR_LG) { - /* User is acessing the DSCR. Set the inherit bit and allow - * the user to set it directly in future by setting via the - * FSCR DSCR bit. We always leave HFSCR DSCR set. + /* + * User is accessing the DSCR register using the problem + * state only SPR number (0x03) either through a mfspr or + * a mtspr instruction. If it is a write attempt through + * a mtspr, then we set the inherit bit. This also allows + * the user to write or read the register directly in the + * future by setting via the FSCR DSCR bit. But in case it + * is a read DSCR attempt through a mfspr instruction, we + * just emulate the instruction instead. This code path will + * always emulate all the mfspr instructions till the user + * has attempted atleast one mtspr instruction. This way it + * preserves the same behaviour when the user is accessing + * the DSCR through privilege level only SPR number (0x11) + * which is emulated through illegal instruction exception. + * We always leave HFSCR DSCR set. */ - current->thread.dscr_inherit = 1; - mtspr(SPRN_FSCR, value | FSCR_DSCR); + if (get_user(instword, (u32 __user *)(regs->nip))) { + pr_err("Failed to fetch the user instruction\n"); + return; + } + + /* Write into DSCR (mtspr 0x03, RS) */ + if ((instword & PPC_INST_MTSPR_DSCR_USER_MASK) + == PPC_INST_MTSPR_DSCR_USER) { + current->thread.dscr_inherit = 1; + mtspr(SPRN_FSCR, value | FSCR_DSCR); + } + + /* Read from DSCR (mfspr RT, 0x03) */ + if ((instword & PPC_INST_MFSPR_DSCR_USER_MASK) + == PPC_INST_MFSPR_DSCR_USER) { + if (emulate_instruction(regs)) { + pr_err("DSCR based mfspr emulation failed\n"); + return; + } + regs->nip += 4; + emulate_single_step(regs); + } return; }
Currently DSCR (Data Stream Control Register) can be accessed with mfspr or mtspr instructions inside a thread via two different SPR numbers. One being the user accessible problem state SPR number 0x03 and the other being the privilege state SPR number 0x11. All access through the privilege state SPR number get emulated through illegal instruction exception. Any access through the problem state SPR number raises one facility unavailable exception which sets the thread based dscr_inherit bit and enables DSCR facility through FSCR register thus allowing direct access to DSCR without going through this exception in the future. We set the thread.dscr_inherit bit whether the access was with mfspr or mtspr instruction which is neither correct nor does it match the behaviour through the instruction emulation code path driven from privilege state SPR number. User currently observes two different kind of behaviour when accessing the DSCR through these two SPR numbers. This problem can be observed through these two test cases by replacing the privilege state SPR number with the problem state SPR number. (1) http://ozlabs.org/~anton/junkcode/dscr_default_test.c (2) http://ozlabs.org/~anton/junkcode/dscr_explicit_test.c This patch fixes the problem by making sure that the behaviour visible to the user remains the same irrespective of which SPR number is being used. Inside facility unavailable exception, we check whether it was cuased by a mfspr or a mtspr isntrucction. In case of mfspr instruction, just emulate the instruction. In case of mtspr instruction, set the thread based dscr_inherit bit and also enable the facility through FSCR. All user SPR based mfspr instruction will be emulated till one user SPR based mtspr has been executed. Signed-off-by: Anshuman Khandual <khandual@linux.vnet.ibm.com> --- arch/powerpc/kernel/traps.c | 43 ++++++++++++++++++++++++++++++++++++++----- 1 file changed, 38 insertions(+), 5 deletions(-)