Re: [PATCH v2 2/4] powerpc/tm: Fix HTM documentation
From: Michael Neuling <hidden>
Date: 2018-08-16 11:57:50
On Mon, 2018-06-18 at 19:59 -0300, Breno Leitao wrote:
This patch simply fix part of the documentation on the HTM code.
=20
This fixes reference to old fields that were renamed in commit
000ec280e3dd ("powerpc: tm: Rename transct_(*) to ck(\1)_state")
=20
It also documents better the flow after commit eb5c3f1c8647 ("powerpc:
Always save/restore checkpointed regs during treclaim/trecheckpoint"),
where tm_recheckpoint can recheckpoint what is in ck{fp,vr}_state blindly=.
=20 Signed-off-by: Breno Leitao <leitao@debian.org>
Acked-By: Michael Neuling <redacted> Thanks
quoted hunk ↗ jump to hunk
--- arch/powerpc/kernel/tm.S | 10 +++++----- arch/powerpc/kernel/traps.c | 15 +++++++++------ 2 files changed, 14 insertions(+), 11 deletions(-) =20diff --git a/arch/powerpc/kernel/tm.S b/arch/powerpc/kernel/tm.S index ff12f47a96b6..019d73053cd3 100644 --- a/arch/powerpc/kernel/tm.S +++ b/arch/powerpc/kernel/tm.S@@ -95,9 +95,9 @@ EXPORT_SYMBOL_GPL(tm_abort); * uint8_t cause) * * - Performs a full reclaim. This destroys outstanding - * transactions and updates thread->regs.tm_ckpt_* with the - * original checkpointed state. Note that thread->regs is - * unchanged. + * transactions and updates thread.ckpt_regs, thread.ckfp_state and + * thread.ckvr_state with the original checkpointed state. Note that + * thread->regs is unchanged. * * Purpose is to both abort transactions of, and preserve the state of, * a transactions at a context switch. We preserve/restore both sets ofprocess@@ -260,7 +260,7 @@ _GLOBAL(tm_reclaim)=20 /* Altivec (VEC/VMX/VR)*/ addi r7, r3, THREAD_CKVRSTATE - SAVE_32VRS(0, r6, r7) /* r6 scratch, r7 transact vr state */ + SAVE_32VRS(0, r6, r7) /* r6 scratch, r7 ckvr_state */ mfvscr v0 li r6, VRSTATE_VSCR stvx v0, r7, r6@@ -271,7 +271,7 @@ _GLOBAL(tm_reclaim)=20 /* Floating Point (FP) */ addi r7, r3, THREAD_CKFPSTATE - SAVE_32FPRS_VSRS(0, R6, R7) /* r6 scratch, r7 transact fp state */ + SAVE_32FPRS_VSRS(0, R6, R7) /* r6 scratch, r7 ckfp_state */ mffs fr0 stfd fr0,FPSTATE_FPSCR(r7) =20diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c index 0e17dcb48720..6742b6b3eb37 100644 --- a/arch/powerpc/kernel/traps.c +++ b/arch/powerpc/kernel/traps.c@@ -1719,16 +1719,19 @@ void fp_unavailable_tm(struct pt_regs *regs) * checkpointed FP registers need to be loaded. */ tm_reclaim_current(TM_CAUSE_FAC_UNAV); - /* Reclaim didn't save out any FPRs to transact_fprs. */ + + /* Reclaim initially saved out bogus (lazy) FPRs to ckfp_state, and + * then it was overwrite by the thr->fp_state by tm_reclaim_thread(). + * + * At this point, ck{fp,vr}_state contains the exact values we want to + * recheckpoint. + */=20 /* Enable FP for the task: */ current->thread.load_fp =3D 1; =20 - /* This loads and recheckpoints the FP registers from - * thread.fpr[]. They will remain in registers after the - * checkpoint so we don't need to reload them after. - * If VMX is in use, the VRs now hold checkpointed values, - * so we don't want to load the VRs from the thread_struct. + /* + * Recheckpoint all the checkpointed ckpt, ck{fp, vr}_state registers. */ tm_recheckpoint(¤t->thread); }