/* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2008 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef sun4v #include #endif #ifndef sun4v #include #endif /* BEGIN CSTYLED */ /* * trapstat: Trap Statistics through Dynamic Trap Table Interposition * ------------------------------------------------------------------- * * Motivation and Overview * * Despite being a fundamental indicator of system behavior, there has * historically been very little insight provided into the frequency and cost * of machine-specific traps. The lack of insight has been especially acute * on UltraSPARC microprocessors: because these microprocessors handle TLB * misses as software traps, the frequency and duration of traps play a * decisive role in the performance of the memory system. As applications have * increasingly outstripped TLB reach, this has become increasingly true. * * Part of the difficulty of observing trap behavior is that the trap handlers * are so frequently called (e.g. millions of times per second) that any * permanently enabled instrumentation would induce an unacceptable performance * degradation. Thus, it is a constraint on any trap observability * infrastructure that it have no probe effect when not explicitly enabled. * * The basic idea, then, is to create an interposing trap table in which each * entry increments a per-trap, in-memory counter and then jumps to the actual, * underlying trap table entry. To enable trapstat, we atomically write to the * trap base address (%tba) register to point to our interposing trap table. * (Note that per-CPU statistics fall out by creating a different trap table * for each CPU.) * * Implementation Details * * While the idea is straight-forward, a nuance of SPARC V9 slightly * complicates the implementation. Unlike its predecessors, SPARC V9 supports * the notion of nested traps. The trap level is kept in the TL register: * during normal operation it is 0; when a trap is taken, the TL register is * incremented by 1. To aid system software, SPARC V9 breaks the trap table * into two halves: the lower half contains the trap handlers for traps taken * when TL is 0; the upper half contains the trap handlers for traps taken * when TL is greater than 0. Each half is further subdivided into two * subsequent halves: the lower half contains the trap handlers for traps * other than those induced by the trap instruction (Tcc variants); the upper * half contains the trap handlers for traps induced by the trap instruction. * This gives a total of four ranges, with each range containing 256 traps: * * +--------------------------------+- 3ff * | | . * | Trap instruction, TL>0 | . * | | . * |- - - - - - - - - - - - - - - - +- 300 * |- - - - - - - - - - - - - - - - +- 2ff * | | . * | Non-trap instruction, TL>0 | . * | | . * |- - - - - - - - - - - - - - - - +- 200 * |- - - - - - - - - - - - - - - - +- 1ff * | | . * | Trap instruction, TL=0 | . * | | . * |- - - - - - - - - - - - - - - - +- 100 * |- - - - - - - - - - - - - - - - +- 0ff * | | . * | Non-trap instruction, TL=0 | . * | | . * +--------------------------------+- 000 * * * Solaris, however, doesn't have reason to support trap instructions when * TL>0 (only privileged code may execute at TL>0; not supporting this only * constrains our own implementation). The trap table actually looks like: * * +--------------------------------+- 2ff * | | . * | Non-trap instruction, TL>0 | . * | | . * |- - - - - - - - - - - - - - - - +- 200 * |- - - - - - - - - - - - - - - - +- 1ff * | | . * | Trap instruction, TL=0 | . * | | . * |- - - - - - - - - - - - - - - - +- 100 * |- - - - - - - - - - - - - - - - +- 0ff * | | . * | Non-trap instruction, TL=0 | . * | | . * +--------------------------------+- 000 * * Putatively to aid system software, SPARC V9 has the notion of multiple * sets of global registers. UltraSPARC defines four sets of global * registers: * * Normal Globals * Alternate Globals (AGs) * MMU Globals (MGs) * Interrupt Globals (IGs) * * The set of globals in use is controlled by bits in PSTATE; when TL is 0 * (and PSTATE has not been otherwise explicitly modified), the Normal Globals * are in use. When a trap is issued, PSTATE is modified to point to a set of * globals corresponding to the trap type. Most traps correspond to the * Alternate Globals, with a minority corresponding to the MMU Globals, and * only the interrupt-vector trap (vector 0x60) corresponding to the Interrupt * Globals. (The complete mapping can be found in the UltraSPARC I&II User's * Manual.) * * Note that the sets of globals are per trap _type_, not per trap _level_. * Thus, when executing a TL>0 trap handler, one may not have registers * available (for example, both trap-instruction traps and spill traps execute * on the alternate globals; if a trap-instruction trap induces a window spill, * the window spill handler has no available globals). For trapstat, this is * problematic: a register is required to transfer control from one arbitrary * location (in the interposing trap table) to another (in the actual trap * table). * * We solve this problem by exploiting the trap table's location at the bottom * of valid kernel memory (i.e. at KERNELBASE). We locate the interposing trap * tables just below KERNELBASE -- thereby allowing us to use a branch-always * instruction (ba) instead of a jump instruction (jmp) to transfer control * from the TL>0 entries in the interposing trap table to the TL>0 entries in * the actual trap table. (N.B. while this allows trap table interposition to * work, it necessarily limits trapstat to only recording information about * TL=0 traps -- there is no way to increment a counter without using a * register.) Diagrammatically: * * Actual trap table: * * +--------------------------------+- 2ff * | | . * | Non-trap instruction, TL>0 | . <-----------------------+ * | | . <-----------------------|-+ * |- - - - - - - - - - - - - - - - +- 200 <-----------------------|-|-+ * |- - - - - - - - - - - - - - - - +- 1ff | | | * | | . | | | * | Trap instruction, TL=0 | . <-----------------+ | | | * | | . <-----------------|-+ | | | * |- - - - - - - - - - - - - - - - +- 100 <-----------------|-|-+ | | | * |- - - - - - - - - - - - - - - - +- 0ff | | | | | | * | | . | | | | | | * | Non-trap instruction, TL=0 | . <-----------+ | | | | | | * | | . <-----------|-+ | | | | | | * +--------------------------------+- 000 <-----------|-|-+ | | | | | | * KERNELBASE | | | | | | | | | * | | | | | | | | | * | | | | | | | | | * Interposing trap table: | | | | | | | | | * | | | | | | | | | * +--------------------------------+- 2ff | | | | | | | | | * | ... | . | | | | | | | | | * | ... | . | | | | | | | | | * | ... | . | | | | | | | | | * |- - - - - - - - - - - - - - - - +- 203 | | | | | | | | | * | ba,a | -------------|-|-|-|-|-|-+ | | * |- - - - - - - - - - - - - - - - +- 202 | | | | | | | | * | ba,a | -------------|-|-|-|-|-|---+ | * |- - - - - - - - - - - - - - - - +- 201 | | | | | | | * | ba,a | -------------|-|-|-|-|-|-----+ * |- - - - - - - - - - - - - - - - +- 200 | | | | | | * | ... | . | | | | | | * | ... | . | | | | | | * | ... | . | | | | | | * |- - - - - - - - - - - - - - - - +- 103 | | | | | | * | (Increment counter) | | | | | | | * | ba,a | -------------------+ | | * |- - - - - - - - - - - - - - - - +- 102 | | | | | * | (Increment counter) | | | | | | * | ba,a | ---------------------+ | * |- - - - - - - - - - - - - - - - +- 101 | | | | * | (Increment counter) | | | | | * | ba,a | -----------------------+ * |- - - - - - - - - - - - - - - - +- 100 | | | * | ... | . | | | * | ... | . | | | * | ... | . | | | * |- - - - - - - - - - - - - - - - +- 003 | | | * | (Increment counter) | | | | * | ba,a | -------------+ | | * |- - - - - - - - - - - - - - - - +- 002 | | * | (Increment counter) | | | * | ba,a | ---------------+ | * |- - - - - - - - - - - - - - - - +- 001 | * | (Increment counter) | | * | ba,a | -----------------+ * +--------------------------------+- 000 * KERNELBASE - tstat_total_size * * tstat_total_size is the number of pages required for each trap table. It * must be true that KERNELBASE - tstat_total_size is less than the maximum * branch displacement; if each CPU were to consume a disjoint virtual range * below KERNELBASE for its trap table, we could support at most * (maximum_branch_displacement / tstat_total_size) CPUs. The maximum branch * displacement for Bicc variants is just under eight megabytes, and (because * the %tba must be 32K aligned), tstat_total_size must be at least 32K; if * each CPU were to consume a disjoint virtual range, we would have an * unacceptably low upper bound of 256 CPUs. * * While there are tricks that one could use to address this constraint (e.g., * creating trampolines every maximum_branch_displacement bytes), we instead * solve this by not permitting each CPU to consume a disjoint virtual range. * Rather, we have each CPU's interposing trap table use the _same_ virtual * range, but we back the trap tables with disjoint physical memory. Normally, * such one-to-many virtual-to-physical mappings are illegal; this is * permissible here only because the pages for the interposing trap table are * necessarily locked in the TLB. (The CPUs thus never have the opportunity to * discover that they have conflicting translations.) * * On CMT architectures in which CPUs can share MMUs, the above trick will not * work: two CPUs that share an MMU cannot have the same virtual address map * to disjoint physical pages. On these architectures, any CPUs sharing the * same MMU must consume a disjoint 32K virtual address range -- limiting the * number of CPUs sharing an MMU on these architectures to 256 due to the * branch displacement limitation described above. On the sun4v architecture, * there is a further limitation: a guest may not have more than eight locked * TLB entries per MMU. To allow operation under this restriction, the * interposing trap table and the trap statistics are each accessed through * a single 4M TLB entry. This limits the footprint to two locked entries * (one for the I-TLB and one for the D-TLB), but further restricts the number * of CPUs to 128 per MMU. However, support for more than 128 CPUs can easily * be added via a hybrid scheme, where the same 4M virtual address is used * on different MMUs. * * On sun4v architecture, we currently don't use hybrid scheme as it imposes * additional restriction on live migration and transparent CPU replacement. * Instead, we increase the number of supported CPUs by reducing the virtual * address space requirements per CPU via shared interposing trap table as * follows: * * Offset (within 4MB page) * +------------------------------------+- 0x400000 * | CPU 507 trap statistics (8KB) | . * |- - - - - - - - - - - - - - - - - - +- 0x3fe000 * | | * | ... | * | | * |- - - - - - - - - - - - - - - - - - +- 0x00c000 * | CPU 1 trap statistics (8KB) | . * |- - - - - - - - - - - - - - - - - - +- 0x00a000 * | CPU 0 trap statistics (8KB) | . * |- - - - - - - - - - - - - - - - - - +- 0x008000 * | Shared trap handler continuation | . * |- - - - - - - - - - - - - - - - - - +- 0x006000 * | Non-trap instruction, TL>0 | . * |- - - - - - - - - - - - - - - - - - +- 0x004000 * | Trap instruction, TL=0 | . * |- - - - - - - - - - - - - - - - - - +- 0x002000 * | Non-trap instruction, TL=0 | . * +------------------------------------+- 0x000000 * * Note that each CPU has its own 8K space for its trap statistics but * shares the same interposing trap handlers. Interposing trap handlers * use the CPU ID to determine the location of per CPU trap statistics * area dynamically. This increases the interposing trap handler overhead, * but is acceptable as it allows us to support up to 508 CPUs with one * 4MB page on sun4v architecture. Support for additional CPUs can be * added via hybrid scheme as mentioned earlier. * * TLB Statistics * * Because TLB misses are an important component of system performance, we wish * to know much more about these traps than simply the number received. * Specifically, we wish to know: * * (a) The amount of time spent executing the TLB miss handler * (b) TLB misses versus TSB misses * (c) Kernel-level misses versus user-level misses * (d) Misses per pagesize * * TLB Statistics: Time Spent Executing * * To accurately determine the amount of time spent executing the TLB miss * handler, one must get a timestamp on trap entry and trap exit, subtract the * latter from the former, and add the result to an accumulating count. * Consider flow of control during normal TLB miss processing (where "ldx * [%g2], %g2" is an arbitrary TLB-missing instruction): * * + - - - - - - - -+ * : : * : ldx [%g2], %g2 :<-------------------------------------------------------+ * : : Return from trap: | * + - - - - - - - -+ TL <- TL - 1 (0) | * | %pc <- TSTATE[TL].TPC (address of load) | * | TLB miss: | * | TL <- TL + 1 (1) | * | %pc <- TLB-miss-trap-handler | * | | * v | * + - - - - - - - - - - - - - - - + | * : : | * : Lookup VA in TSB : | * : If (hit) : | * : Fill TLB : | * : Else : | * : Lookup VA (hme hash table : | * : or segkpm) : | * : Fill TLB : | * : Endif : | * : Issue "retry" ---------------------------------------------------------+ * : : * + - - - - - - - - - - - - - - - + * TLB-miss-trap-handler * * * As the above diagram indicates, interposing on the trap table allows one * only to determine a timestamp on trap _entry_: when the TLB miss handler * has completed filling the TLB, a "retry" will be issued, and control will * transfer immediately back to the missing %pc. * * To obtain a timestamp on trap exit, we must then somehow interpose between * the "retry" and the subsequent control transfer to the TLB-missing * instruction. To do this, we _push_ a trap level. The basic idea is to * spoof a TLB miss by raising TL, setting the %tpc to be within text * controlled by trapstat (the "TLB return entry") and branching to the * underlying TLB miss handler. When the TLB miss handler issues its "retry", * control will transfer not to the TLB-missing instruction, but rather to the * TLB return entry. This code can then obtain a timestamp, and issue its own * "retry" -- thereby correctly returning to the TLB-missing instruction. * Here is the above TLB miss flow control diagram modified to reflect * trapstat's operation: * * + - - - - - - - -+ * : : * : ldx [%g2], %g2 :<-------------------------------------------------------+ * : : Return from trap: | * + - - - - - - - -+ TL <- TL - 1 (0) | * | %pc <- TSTATE[TL].TPC (address of load) | * | TLB miss: | * | TL <- TL + 1 (1) | * | %pc <- TLB-miss-trap-handler (trapstat) | * | | * v TLB-return-entry (trapstat) | * + - - - - - - - - - - - - - - - - - - + + - - - - - - - - - - - - - + | * : : : : | * : Record timestamp : : Record timestamp : | * : TL <- 2 : : Take timestamp difference : | * : TSTATE[1].TPC <- TLB-return-entry : : Add to running total : | * : ba,a TLB-miss-trap-handler -----------+ : Issue "retry" --------------+ * : : | : : * + - - - - - - - - - - - - - - - - - - + | + - - - - - - - - - - - - - + * TLB-miss-trap-handler | ^ * (trapstat) | | * | | * | | * +-----------------------+ | * | | * | | * v | * + - - - - - - - - - - - - - - - + | * : : | * : Lookup VA in TSB : | * : If (hit) : | * : Fill TLB : | * : Else : | * : Lookup VA (hme hash table : | * : or segkpm) : | * : Fill TLB : | * : Endif : | * : Issue "retry" ------------------------------------------+ * : : Return from trap: * + - - - - - - - - - - - - - - - + TL <- TL - 1 (1) * TLB-miss-trap-handler %pc <- TSTATE[TL].TPC (TLB-return-entry) * * * A final subterfuge is required to complete our artifice: if we miss in * the TLB, the TSB _and_ the subsequent hash or segkpm lookup (that is, if * there is no valid translation for the TLB-missing address), common system * software will need to accurately determine the %tpc as part of its page * fault handling. We therefore modify the kernel to check the %tpc in this * case: if the %tpc falls within the VA range controlled by trapstat and * the TL is 2, TL is simply lowered back to 1 (this check is implemented * by the TSTAT_CHECK_TL1 macro). Lowering TL to 1 has the effect of * discarding the state pushed by trapstat. * * TLB Statistics: TLB Misses versus TSB Misses * * Distinguishing TLB misses from TSB misses requires further interposition * on the TLB miss handler: we cannot know a priori or a posteriori if a * given VA will or has hit in the TSB. * * We achieve this distinction by adding a second TLB return entry almost * identical to the first -- differing only in the address to which it * stores its results. We then modify the TLB miss handlers of the kernel * such that they check the %tpc when they determine that a TLB miss has * subsequently missed in the TSB: if the %tpc lies within trapstat's VA * range and TL is 2 (that is, if trapstat is running), the TLB miss handler * _increments_ the %tpc by the size of the TLB return entry. The ensuing * "retry" will thus transfer control to the second TLB return entry, and * the time spent in the handler will be accumulated in a memory location * specific to TSB misses. * * N.B.: To minimize the amount of knowledge the kernel must have of trapstat, * we do not allow the kernel to hard-code the size of the TLB return entry. * Rather, the actual tsbmiss handler executes a known instruction at the * corresponding tsbmiss patch points (see the tstat_tsbmiss_patch_table) with * the %tpc in %g7: when trapstat is not running, these points contain the * harmless TSTAT_TSBMISS_INSTR instruction ("add %g7, 0, %g7"). Before * running, trapstat modifies the instructions at these patch points such * that the simm13 equals the size of the TLB return entry. * * TLB Statistics: Kernel-level Misses versus User-level Misses * * Differentiating user-level misses from kernel-level misses employs a * similar technique, but is simplified by the ability to distinguish a * user-level miss from a kernel-level miss a priori by reading the context * register: we implement kernel-/user-level differentiation by again doubling * the number of TLB return entries, and setting the %tpc to the appropriate * TLB return entry in trapstat's TLB miss handler. Together with the doubling * of entries required for TLB-miss/TSB-miss differentiation, this yields a * total of four TLB return entries: * * Level TSB hit? Structure member * ------------------------------------------------------------ * Kernel Yes tstat_tlbret_t.ttlbr_ktlb * Kernel No tstat_tlbret_t.ttlbr_ktsb * User Yes tstat_tlbret_t.ttlbr_utlb * User No tstat_tlbret_t.ttlbr_utsb * * TLB Statistics: Misses per Pagesize * * As with the TLB-/TSB-miss differentiation, we have no way of determining * pagesize a priori. This is therefore implemented by mandating a new rule: * whenever the kernel fills the TLB in its TLB miss handler, the TTE * corresponding to the TLB-missing VA must be in %g5 when the handler * executes its "retry". This allows the TLB return entry to determine * pagesize by simply looking at the pagesize field in the TTE stored in * %g5. * * TLB Statistics: Probe Effect * * As one might imagine, gathering TLB statistics by pushing a trap level * induces significant probe effect. To account for this probe effect, * trapstat attempts to observe it by executing a code sequence with a known * number of TLB misses both before and after interposing on the trap table. * This allows trapstat to determine a per-trap probe effect which can then be * factored into the "%tim" fields of the trapstat command. * * Note that on sun4v platforms, TLB misses are normally handled by the * hypervisor or the hardware TSB walker. Thus no fast MMU miss information * is reported for normal operation. However, when trapstat is invoked * with -t or -T option to collect detailed TLB statistics, kernel takes * over TLB miss handling. This results in significantly more overhead * and TLB statistics may not be as accurate as on sun4u platforms. * On some processors, hypervisor or hardware may provide a low overhead * interface to collect TSB hit statistics. This support is exposed via * a well defined CPU module interface (cpu_trapstat_conf to enable this * interface and cpu_trapstat_data to get detailed TSB hit statistics). * In this scenario, TSB miss statistics is collected by intercepting the * IMMU_miss and DMMU_miss traps using above mentioned trap interposition * approach. * * Locking * * The implementation uses two locks: tstat_lock (a local lock) and the global * cpu_lock. tstat_lock is used to assure trapstat's consistency in the * presence of multithreaded /dev/trapstat consumers (while as of this writing * the only consumer of /dev/trapstat is single threaded, it is obviously * necessary to correctly support multithreaded access). cpu_lock is held * whenever CPUs are being manipulated directly, to prevent them from * disappearing in the process. Because trapstat's DR callback * (trapstat_cpu_setup()) must grab tstat_lock and is called with cpu_lock * held, the lock ordering is necessarily cpu_lock before tstat_lock. * */ /* END CSTYLED */ static dev_info_t *tstat_devi; /* saved in xxattach() for xxinfo() */ static int tstat_open; /* set if driver is open */ static kmutex_t tstat_lock; /* serialize access */ static vmem_t *tstat_arena; /* arena for TLB-locked pages */ static tstat_percpu_t *tstat_percpu; /* per-CPU data */ static int tstat_running; /* set if trapstat is running */ static tstat_data_t *tstat_buffer; /* staging buffer for outgoing data */ static int tstat_options; /* bit-wise indication of options */ static int *tstat_enabled; /* map of enabled trap entries */ static int tstat_tsbmiss_patched; /* tsbmiss patch flag */ static callb_id_t tstat_cprcb; /* CPR callback */ static char *tstat_probe_area; /* VA range used for probe effect */ static caddr_t tstat_probe_phys; /* physical to back above VA */ static hrtime_t tstat_probe_time; /* time spent on probe effect */ static hrtime_t tstat_probe_before[TSTAT_PROBE_NLAPS]; static hrtime_t tstat_probe_after[TSTAT_PROBE_NLAPS]; static uint_t tstat_pgszs; /* # of kernel page sizes */ static uint_t tstat_user_pgszs; /* # of user page sizes */ /* * sizeof tstat_data_t + pgsz data for the kernel. For simplicity's sake, when * we collect data, we do it based upon szc, but when we report data back to * userland, we have to do it based upon the userszc which may not match. * So, these two variables are for internal use and exported use respectively. */ static size_t tstat_data_t_size; static size_t tstat_data_t_exported_size; #ifndef sun4v static size_t tstat_data_pages; /* number of pages of tstat data */ static size_t tstat_data_size; /* tstat data size in bytes */ static size_t tstat_total_pages; /* #data pages + #instr pages */ static size_t tstat_total_size; /* tstat data size + instr size */ #else /* sun4v */ static caddr_t tstat_va; /* VA of memory reserved for TBA */ static pfn_t tstat_pfn; /* PFN of memory reserved for TBA */ static boolean_t tstat_fast_tlbstat = B_FALSE; static int tstat_traptab_initialized; #endif /* sun4v */ /* * In the above block comment, see "TLB Statistics: TLB Misses versus * TSB Misses" for an explanation of the tsbmiss patch points. */ extern uint32_t tsbmiss_trapstat_patch_point; extern uint32_t tsbmiss_trapstat_patch_point_kpm; extern uint32_t tsbmiss_trapstat_patch_point_kpm_small; /* * Trapstat tsbmiss patch table */ tstat_tsbmiss_patch_entry_t tstat_tsbmiss_patch_table[] = { {(uint32_t *)&tsbmiss_trapstat_patch_point, 0}, {(uint32_t *)&tsbmiss_trapstat_patch_point_kpm, 0}, {(uint32_t *)&tsbmiss_trapstat_patch_point_kpm_small, 0}, {(uint32_t *)NULL, 0} }; /* * We define some general SPARC-specific constants to allow more readable * relocations. */ #define NOP 0x01000000 #define HI22(v) ((uint32_t)(v) >> 10) #define LO10(v) ((uint32_t)(v) & 0x3ff) #define LO12(v) ((uint32_t)(v) & 0xfff) #define DISP22(from, to) \ ((((uintptr_t)(to) - (uintptr_t)(from)) >> 2) & 0x3fffff) #define ASI(asi) ((asi) << 5) /* * The interposing trap table must be locked in the I-TLB, and any data * referred to in the interposing trap handler must be locked in the D-TLB. * This function locks these pages in the appropriate TLBs by creating TTEs * from whole cloth, and manually loading them into the TLB. This function is * called from cross call context. * * On sun4v platforms, we use 4M page size mappings to minimize the number * of locked down entries (i.e. permanent mappings). Each CPU uses a * reserved portion of that 4M page for its TBA and data. */ static void trapstat_load_tlb(void) { #ifndef sun4v int i; #else uint64_t ret; #endif tte_t tte; tstat_percpu_t *tcpu = &tstat_percpu[CPU->cpu_id]; caddr_t va = tcpu->tcpu_vabase; ASSERT(tcpu->tcpu_flags & TSTAT_CPU_ALLOCATED); ASSERT(!(tcpu->tcpu_flags & TSTAT_CPU_ENABLED)); #ifndef sun4v for (i = 0; i < tstat_total_pages; i++, va += MMU_PAGESIZE) { tte.tte_inthi = TTE_VALID_INT | TTE_SZ_INT(TTE8K) | TTE_PFN_INTHI(tcpu->tcpu_pfn[i]); if (i < TSTAT_INSTR_PAGES) { tte.tte_intlo = TTE_PFN_INTLO(tcpu->tcpu_pfn[i]) | TTE_LCK_INT | TTE_CP_INT | TTE_PRIV_INT; sfmmu_itlb_ld_kva(va, &tte); } else { tte.tte_intlo = TTE_PFN_INTLO(tcpu->tcpu_pfn[i]) | TTE_LCK_INT | TTE_CP_INT | TTE_CV_INT | TTE_PRIV_INT | TTE_HWWR_INT; sfmmu_dtlb_ld_kva(va, &tte); } } #else /* sun4v */ tte.tte_inthi = TTE_VALID_INT | TTE_PFN_INTHI(tstat_pfn); tte.tte_intlo = TTE_PFN_INTLO(tstat_pfn) | TTE_CP_INT | TTE_CV_INT | TTE_PRIV_INT | TTE_HWWR_INT | TTE_SZ_INTLO(TTE4M); ret = hv_mmu_map_perm_addr(va, KCONTEXT, *(uint64_t *)&tte, MAP_ITLB | MAP_DTLB); if (ret != H_EOK) cmn_err(CE_PANIC, "trapstat: cannot map new TBA " "for cpu %d (error: 0x%lx)", CPU->cpu_id, ret); #endif /* sun4v */ } /* * As mentioned in the "TLB Statistics: TLB Misses versus TSB Misses" section * of the block comment, TLB misses are differentiated from TSB misses in * part by hot-patching the instructions at the tsbmiss patch points (see * tstat_tsbmiss_patch_table). This routine is used both to initially patch * the instructions, and to patch them back to their original values upon * restoring the original trap table. */ static void trapstat_hotpatch() { uint32_t instr; uint32_t simm13; tstat_tsbmiss_patch_entry_t *ep; ASSERT(MUTEX_HELD(&tstat_lock)); if (!(tstat_options & TSTAT_OPT_TLBDATA)) return; if (!tstat_tsbmiss_patched) { /* * We haven't patched the TSB paths; do so now. */ /*CONSTCOND*/ ASSERT(offsetof(tstat_tlbret_t, ttlbr_ktsb) - offsetof(tstat_tlbret_t, ttlbr_ktlb) == offsetof(tstat_tlbret_t, ttlbr_utsb) - offsetof(tstat_tlbret_t, ttlbr_utlb)); simm13 = offsetof(tstat_tlbret_t, ttlbr_ktsb) - offsetof(tstat_tlbret_t, ttlbr_ktlb); for (ep = tstat_tsbmiss_patch_table; ep->tpe_addr; ep++) { ASSERT(ep->tpe_instr == 0); instr = ep->tpe_instr = *ep->tpe_addr; /* * Assert that the instruction we're about to patch is * "add %g7, 0, %g7" (0x8e01e000). */ ASSERT(instr == TSTAT_TSBMISS_INSTR); instr |= simm13; hot_patch_kernel_text((caddr_t)ep->tpe_addr, instr, sizeof (instr)); } tstat_tsbmiss_patched = 1; } else { /* * Remove patches from the TSB paths. */ for (ep = tstat_tsbmiss_patch_table; ep->tpe_addr; ep++) { ASSERT(ep->tpe_instr == TSTAT_TSBMISS_INSTR); hot_patch_kernel_text((caddr_t)ep->tpe_addr, ep->tpe_instr, sizeof (instr)); ep->tpe_instr = 0; } tstat_tsbmiss_patched = 0; } } /* * This is the routine executed to clock the performance of the trap table, * executed both before and after interposing on the trap table to attempt to * determine probe effect. The probe effect is used to adjust the "%tim" * fields of trapstat's -t and -T output; we only use TLB misses to clock the * trap table. We execute the inner loop (which is designed to exceed the * TLB's reach) nlaps times, taking the best time as our time (thereby * factoring out the effects of interrupts, cache misses or other perturbing * events. */ static hrtime_t trapstat_probe_laps(int nlaps, hrtime_t *buf) { int i, j = 0; hrtime_t ts, best = INT64_MAX; while (nlaps--) { ts = rdtick(); for (i = 0; i < TSTAT_PROBE_SIZE; i += MMU_PAGESIZE) *((volatile char *)&tstat_probe_area[i]); if ((ts = rdtick() - ts) < best) best = ts; buf[j++] = ts; } return (best); } /* * This routine determines the probe effect by calling trapstat_probe_laps() * both without and with the interposing trap table. Note that this is * called from a cross call on the desired CPU, and that it is called on * every CPU (this is necessary because the probe effect may differ from * one CPU to another). */ static void trapstat_probe() { tstat_percpu_t *tcpu = &tstat_percpu[CPU->cpu_id]; hrtime_t before, after; if (!(tcpu->tcpu_flags & TSTAT_CPU_SELECTED)) return; if (tstat_probe_area == NULL || (tstat_options & TSTAT_OPT_NOGO)) return; /* * We very much expect the %tba to be KERNELBASE; this is a * precautionary measure to assure that trapstat doesn't melt the * machine should the %tba point unexpectedly elsewhere. */ if (get_tba() != (caddr_t)KERNELBASE) return; /* * Preserve this CPU's data before destroying it by enabling the * interposing trap table. We can safely use tstat_buffer because * the caller of the trapstat_probe() cross call is holding tstat_lock. */ bcopy(tcpu->tcpu_data, tstat_buffer, tstat_data_t_size); tstat_probe_time = gethrtime(); before = trapstat_probe_laps(TSTAT_PROBE_NLAPS, tstat_probe_before); (void) set_tba(tcpu->tcpu_ibase); after = trapstat_probe_laps(TSTAT_PROBE_NLAPS, tstat_probe_after); (void) set_tba((caddr_t)KERNELBASE); tstat_probe_time = gethrtime() - tstat_probe_time; bcopy(tstat_buffer, tcpu->tcpu_data, tstat_data_t_size); tcpu->tcpu_data->tdata_peffect = (after - before) / TSTAT_PROBE_NPAGES; } static void trapstat_probe_alloc() { pfn_t pfn; caddr_t va; int i; ASSERT(MUTEX_HELD(&tstat_lock)); ASSERT(tstat_probe_area == NULL); ASSERT(tstat_probe_phys == NULL); if (!(tstat_options & TSTAT_OPT_TLBDATA)) return; /* * Grab some virtual from the heap arena. */ tstat_probe_area = vmem_alloc(heap_arena, TSTAT_PROBE_SIZE, VM_SLEEP); va = tstat_probe_area; /* * Grab a single physical page. */ tstat_probe_phys = vmem_alloc(tstat_arena, MMU_PAGESIZE, VM_SLEEP); pfn = hat_getpfnum(kas.a_hat, tstat_probe_phys); /* * Now set the translation for every page in our virtual range * to be our allocated physical page. */ for (i = 0; i < TSTAT_PROBE_NPAGES; i++) { hat_devload(kas.a_hat, va, MMU_PAGESIZE, pfn, PROT_READ, HAT_LOAD_NOCONSIST | HAT_LOAD_LOCK); va += MMU_PAGESIZE; } } static void trapstat_probe_free() { caddr_t va; int i; ASSERT(MUTEX_HELD(&tstat_lock)); if ((va = tstat_probe_area) == NULL) return; for (i = 0; i < TSTAT_PROBE_NPAGES; i++) { hat_unload(kas.a_hat, va, MMU_PAGESIZE, HAT_UNLOAD_UNLOCK); va += MMU_PAGESIZE; } vmem_free(tstat_arena, tstat_probe_phys, MMU_PAGESIZE); vmem_free(heap_arena, tstat_probe_area, TSTAT_PROBE_SIZE); tstat_probe_phys = NULL; tstat_probe_area = NULL; } /* * This routine actually enables a CPU by setting its %tba to be the * CPU's interposing trap table. It is called out of cross call context. */ static void trapstat_enable() { tstat_percpu_t *tcpu = &tstat_percpu[CPU->cpu_id]; if (!(tcpu->tcpu_flags & TSTAT_CPU_SELECTED)) return; ASSERT(tcpu->tcpu_flags & TSTAT_CPU_ALLOCATED); ASSERT(!(tcpu->tcpu_flags & TSTAT_CPU_ENABLED)); if (get_tba() != (caddr_t)KERNELBASE) return; if (!(tstat_options & TSTAT_OPT_NOGO)) (void) set_tba(tcpu->tcpu_ibase); tcpu->tcpu_flags |= TSTAT_CPU_ENABLED; #ifdef sun4v if ((tstat_options & TSTAT_OPT_TLBDATA) && !(tstat_options & TSTAT_OPT_NOGO)) { if (tstat_fast_tlbstat) { /* * Invoke processor specific interface to enable * collection of TSB hit statistics. */ cpu_trapstat_conf(CPU_TSTATCONF_ENABLE); } else { /* * Collect TLB miss statistics by taking over * TLB miss handling from the hypervisor. This * is done by telling the hypervisor that there * is no TSB configured. Also set TSTAT_TLB_STATS * flag so that no user TSB is configured during * context switch time. */ cpu_t *cp = CPU; cp->cpu_m.cpu_tstat_flags |= TSTAT_TLB_STATS; (void) hv_set_ctx0(NULL, NULL); (void) hv_set_ctxnon0(NULL, NULL); } } #endif } /* * This routine disables a CPU (vis a vis trapstat) by setting its %tba to be * the actual, underlying trap table. It is called out of cross call context. */ static void trapstat_disable() { tstat_percpu_t *tcpu = &tstat_percpu[CPU->cpu_id]; if (!(tcpu->tcpu_flags & TSTAT_CPU_ENABLED)) return; ASSERT(tcpu->tcpu_flags & TSTAT_CPU_SELECTED); ASSERT(tcpu->tcpu_flags & TSTAT_CPU_ALLOCATED); if (!(tstat_options & TSTAT_OPT_NOGO)) (void) set_tba((caddr_t)KERNELBASE); tcpu->tcpu_flags &= ~TSTAT_CPU_ENABLED; #ifdef sun4v if ((tstat_options & TSTAT_OPT_TLBDATA) && !(tstat_options & TSTAT_OPT_NOGO)) { if (tstat_fast_tlbstat) { /* * Invoke processor specific interface to disable * collection of TSB hit statistics on each processor. */ cpu_trapstat_conf(CPU_TSTATCONF_DISABLE); } else { /* * As part of collecting TLB miss statistics, we took * over TLB miss handling from the hypervisor by * telling the hypervisor that NO TSB is configured. * We need to restore that by communicating proper * kernel/user TSB information so that TLB misses * can be handled by the hypervisor or the hardware * more efficiently. * * We restore kernel TSB information right away. * However, to minimize any locking dependency, we * don't restore user TSB information right away. * Instead, we simply clear the TSTAT_TLB_STATS flag * so that the user TSB information is automatically * restored on next context switch. * * Note that the call to restore kernel TSB information * will normally not fail, unless wrong information is * passed here. In that scenario, system will still * continue to function properly with the exception of * kernel handling all the TLB misses. */ struct hv_tsb_block *hvbp = &ksfmmup->sfmmu_hvblock; cpu_t *cp = CPU; cp->cpu_m.cpu_tstat_flags &= ~TSTAT_TLB_STATS; (void) hv_set_ctx0(hvbp->hv_tsb_info_cnt, hvbp->hv_tsb_info_pa); } } #endif } /* * We use %tick as the time base when recording the time spent executing * the trap handler. %tick, however, is not necessarily kept in sync * across CPUs (indeed, different CPUs may have different %tick frequencies). * We therefore cross call onto a CPU to get a snapshot of its data to * copy out; this is the routine executed out of that cross call. */ static void trapstat_snapshot() { tstat_percpu_t *tcpu = &tstat_percpu[CPU->cpu_id]; tstat_data_t *data = tcpu->tcpu_data; ASSERT(tcpu->tcpu_flags & TSTAT_CPU_SELECTED); ASSERT(tcpu->tcpu_flags & TSTAT_CPU_ALLOCATED); ASSERT(tcpu->tcpu_flags & TSTAT_CPU_ENABLED); data->tdata_snapts = gethrtime(); data->tdata_snaptick = rdtick(); bcopy(data, tstat_buffer, tstat_data_t_size); #ifdef sun4v /* * Invoke processor specific interface to collect TSB hit * statistics on each processor. */ if ((tstat_options & TSTAT_OPT_TLBDATA) && tstat_fast_tlbstat) cpu_trapstat_data((void *) tstat_buffer->tdata_pgsz, tstat_pgszs); #endif } /* * The TSTAT_RETENT_* constants define offsets in the TLB return entry. * They are used only in trapstat_tlbretent() (below) and #undef'd * immediately afterwards. Any change to "retent" in trapstat_tlbretent() * will likely require changes to these constants. */ #ifndef sun4v #define TSTAT_RETENT_STATHI 1 #define TSTAT_RETENT_STATLO 2 #define TSTAT_RETENT_SHIFT 11 #define TSTAT_RETENT_COUNT_LD 13 #define TSTAT_RETENT_COUNT_ST 15 #define TSTAT_RETENT_TMPTSHI 16 #define TSTAT_RETENT_TMPTSLO 17 #define TSTAT_RETENT_TIME_LD 19 #define TSTAT_RETENT_TIME_ST 21 #else /* sun4v */ #define TSTAT_RETENT_TDATASHFT 2 #define TSTAT_RETENT_STATHI 4 #define TSTAT_RETENT_STATLO 6 #define TSTAT_RETENT_SHIFT 9 #define TSTAT_RETENT_COUNT_LD 11 #define TSTAT_RETENT_COUNT_ST 13 #define TSTAT_RETENT_TMPTSHI 14 #define TSTAT_RETENT_TMPTSLO 16 #define TSTAT_RETENT_TIME_LD 18 #define TSTAT_RETENT_TIME_ST 20 #endif /* sun4v */ static void trapstat_tlbretent(tstat_percpu_t *tcpu, tstat_tlbretent_t *ret, tstat_missdata_t *data) { uint32_t *ent = ret->ttlbrent_instr, shift; uintptr_t base; #ifndef sun4v uintptr_t tmptick = TSTAT_DATA_OFFS(tcpu, tdata_tmptick); #else uintptr_t tmptick = TSTAT_CPU0_DATA_OFFS(tcpu, tdata_tmptick); #endif /* * This is the entry executed upon return from the TLB/TSB miss * handler (i.e. the code interpositioned between the "retry" and * the actual return to the TLB-missing instruction). Detail on its * theory of operation can be found in the "TLB Statistics" section * of the block comment. Note that we expect the TTE just loaded * into the TLB to be in %g5; all other globals are available as * scratch. Finally, note that the page size information in sun4v is * located in the lower bits of the TTE -- requiring us to have a * different return entry on sun4v. */ static const uint32_t retent[TSTAT_TLBRET_NINSTR] = { #ifndef sun4v 0x87410000, /* rd %tick, %g3 */ 0x03000000, /* sethi %hi(stat), %g1 */ 0x82106000, /* or %g1, %lo(stat), %g1 */ 0x89297001, /* sllx %g5, 1, %g4 */ 0x8931303e, /* srlx %g4, 62, %g4 */ 0x8531702e, /* srlx %g5, 46, %g2 */ 0x8408a004, /* and %g2, 4, %g2 */ 0x88110002, /* or %g4, %g2, %g4 */ 0x80a12005, /* cmp %g4, 5 */ 0x34400002, /* bg,a,pn %icc, +8 */ 0x88102004, /* mov 4, %g4 */ 0x89292000, /* sll %g4, shift, %g4 */ 0x82004004, /* add %g1, %g4, %g1 */ 0xc4586000, /* ldx [%g1 + tmiss_count], %g2 */ 0x8400a001, /* add %g2, 1, %g2 */ 0xc4706000, /* stx %g2, [%g1 + tmiss_count] */ 0x0d000000, /* sethi %hi(tdata_tmptick), %g6 */ 0xc459a000, /* ldx [%g6 + %lo(tdata_tmptick)], %g2 */ 0x8620c002, /* sub %g3, %g2, %g3 */ 0xc4586000, /* ldx [%g1 + tmiss_time], %g2 */ 0x84008003, /* add %g2, %g3, %g2 */ 0xc4706000, /* stx %g2, [%g1 + tmiss_time] */ 0x83f00000 /* retry */ #else /* sun4v */ 0x82102008, /* mov SCRATCHPAD_CPUID, %g1 */ 0xced84400, /* ldxa [%g1]ASI_SCRATCHPAD, %g7 */ 0x8f29f000, /* sllx %g7, TSTAT_DATA_SHIFT, %g7 */ 0x87410000, /* rd %tick, %g3 */ 0x03000000, /* sethi %hi(stat), %g1 */ 0x82004007, /* add %g1, %g7, %g1 */ 0x82106000, /* or %g1, %lo(stat), %g1 */ 0x8929703d, /* sllx %g5, 61, %g4 */ 0x8931303d, /* srlx %g4, 61, %g4 */ 0x89292000, /* sll %g4, shift, %g4 */ 0x82004004, /* add %g1, %g4, %g1 */ 0xc4586000, /* ldx [%g1 + tmiss_count], %g2 */ 0x8400a001, /* add %g2, 1, %g2 */ 0xc4706000, /* stx %g2, [%g1 + tmiss_count] */ 0x0d000000, /* sethi %hi(tdata_tmptick), %g6 */ 0x8c018007, /* add %g6, %g7, %g6 */ 0xc459a000, /* ldx [%g6 + %lo(tdata_tmptick)], %g2 */ 0x8620c002, /* sub %g3, %g2, %g3 */ 0xc4586000, /* ldx [%g1 + tmiss_time], %g2 */ 0x84008003, /* add %g2, %g3, %g2 */ 0xc4706000, /* stx %g2, [%g1 + tmiss_time] */ 0x83f00000 /* retry */ #endif /* sun4v */ }; ASSERT(MUTEX_HELD(&tstat_lock)); /*CONSTCOND*/ ASSERT(offsetof(tstat_missdata_t, tmiss_count) <= LO10(-1)); /*CONSTCOND*/ ASSERT(offsetof(tstat_missdata_t, tmiss_time) <= LO10(-1)); /*CONSTCOND*/ ASSERT(!((sizeof (tstat_pgszdata_t) - 1) & sizeof (tstat_pgszdata_t))); for (shift = 1; (1 << shift) != sizeof (tstat_pgszdata_t); shift++) continue; base = (uintptr_t)tcpu->tcpu_ibase + TSTAT_INSTR_SIZE + ((uintptr_t)data - (uintptr_t)tcpu->tcpu_data); bcopy(retent, ent, sizeof (retent)); #if defined(sun4v) ent[TSTAT_RETENT_TDATASHFT] |= LO10((uintptr_t)TSTAT_DATA_SHIFT); #endif ent[TSTAT_RETENT_STATHI] |= HI22(base); ent[TSTAT_RETENT_STATLO] |= LO10(base); ent[TSTAT_RETENT_SHIFT] |= shift; /* LINTED E_EXPR_NULL_EFFECT */ ent[TSTAT_RETENT_COUNT_LD] |= offsetof(tstat_missdata_t, tmiss_count); /* LINTED E_EXPR_NULL_EFFECT */ ent[TSTAT_RETENT_COUNT_ST] |= offsetof(tstat_missdata_t, tmiss_count); ent[TSTAT_RETENT_TMPTSHI] |= HI22(tmptick); ent[TSTAT_RETENT_TMPTSLO] |= LO10(tmptick); ent[TSTAT_RETENT_TIME_LD] |= offsetof(tstat_missdata_t, tmiss_time); ent[TSTAT_RETENT_TIME_ST] |= offsetof(tstat_missdata_t, tmiss_time); } #if defined(sun4v) #undef TSTAT_RETENT_TDATASHFT #endif #undef TSTAT_RETENT_STATHI #undef TSTAT_RETENT_STATLO #undef TSTAT_RETENT_SHIFT #undef TSTAT_RETENT_COUNT_LD #undef TSTAT_RETENT_COUNT_ST #undef TSTAT_RETENT_TMPTSHI #undef TSTAT_RETENT_TMPTSLO #undef TSTAT_RETENT_TIME_LD #undef TSTAT_RETENT_TIME_ST /* * The TSTAT_TLBENT_* constants define offsets in the TLB entry. They are * used only in trapstat_tlbent() (below) and #undef'd immediately afterwards. * Any change to "tlbent" in trapstat_tlbent() will likely require changes * to these constants. */ #ifndef sun4v #define TSTAT_TLBENT_STATHI 0 #define TSTAT_TLBENT_STATLO_LD 1 #define TSTAT_TLBENT_STATLO_ST 3 #define TSTAT_TLBENT_MMUASI 15 #define TSTAT_TLBENT_TPCHI 18 #define TSTAT_TLBENT_TPCLO_USER 19 #define TSTAT_TLBENT_TPCLO_KERN 21 #define TSTAT_TLBENT_TSHI 25 #define TSTAT_TLBENT_TSLO 27 #define TSTAT_TLBENT_BA 28 #else /* sun4v */ #define TSTAT_TLBENT_TDATASHFT 2 #define TSTAT_TLBENT_STATHI 3 #define TSTAT_TLBENT_STATLO_LD 5 #define TSTAT_TLBENT_STATLO_ST 7 #define TSTAT_TLBENT_TAGTARGET 23 #define TSTAT_TLBENT_TPCHI 25 #define TSTAT_TLBENT_TPCLO_USER 26 #define TSTAT_TLBENT_TPCLO_KERN 28 #define TSTAT_TLBENT_TSHI 32 #define TSTAT_TLBENT_TSLO 35 #define TSTAT_TLBENT_BA 36 #endif /* sun4v */ static void trapstat_tlbent(tstat_percpu_t *tcpu, int entno) { uint32_t *ent; uintptr_t orig, va, baoffs; #ifndef sun4v int itlb = entno == TSTAT_ENT_ITLBMISS; uint32_t asi = itlb ? ASI(ASI_IMMU) : ASI(ASI_DMMU); #else int itlb = (entno == TSTAT_ENT_IMMUMISS || entno == TSTAT_ENT_ITLBMISS); uint32_t tagtarget_off = itlb ? MMFSA_I_CTX : MMFSA_D_CTX; uint32_t *tent; /* MMU trap vector entry */ uintptr_t tentva; /* MMU trap vector entry va */ static const uint32_t mmumiss[TSTAT_ENT_NINSTR] = { 0x30800000, /* ba,a addr */ NOP, NOP, NOP, NOP, NOP, NOP, NOP }; #endif int entoffs = entno << TSTAT_ENT_SHIFT; uintptr_t tmptick, stat, tpc, utpc; tstat_pgszdata_t *data = &tcpu->tcpu_data->tdata_pgsz[0]; tstat_tlbdata_t *udata, *kdata; tstat_tlbret_t *ret; /* * When trapstat is run with TLB statistics, this is the entry for * both I- and D-TLB misses; this code performs trap level pushing, * as described in the "TLB Statistics" section of the block comment. * This code is executing at TL 1; %tstate[0] contains the saved * state at the time of the TLB miss. Pushing trap level 1 (and thus * raising TL to 2) requires us to fill in %tstate[1] with our %pstate, * %cwp and %asi. We leave %tt unchanged, and we set %tpc and %tnpc to * the appropriate TLB return entry (based on the context of the miss). * Finally, we sample %tick, and stash it in the tdata_tmptick member * the per-CPU tstat_data structure. tdata_tmptick will be used in * the TLB return entry to determine the amount of time spent in the * TLB miss handler. * * Note that on sun4v platforms, we must obtain the context information * from the MMU fault status area. (The base address of this MMU fault * status area is kept in the scratchpad register 0.) */ static const uint32_t tlbent[] = { #ifndef sun4v 0x03000000, /* sethi %hi(stat), %g1 */ 0xc4586000, /* ldx [%g1 + %lo(stat)], %g2 */ 0x8400a001, /* add %g2, 1, %g2 */ 0xc4706000, /* stx %g2, [%g1 + %lo(stat)] */ 0x85524000, /* rdpr %cwp, %g2 */ 0x87518000, /* rdpr %pstate, %g3 */ 0x8728f008, /* sllx %g3, 8, %g3 */ 0x84108003, /* or %g2, %g3, %g2 */ 0x8740c000, /* rd %asi, %g3 */ 0x8728f018, /* sllx %g3, 24, %g3 */ 0x84108003, /* or %g2, %g3, %g2 */ 0x8350c000, /* rdpr %tt, %g1 */ 0x8f902002, /* wrpr %g0, 2, %tl */ 0x85908000, /* wrpr %g2, %g0, %tstate */ 0x87904000, /* wrpr %g1, %g0, %tt */ 0xc2d80000, /* ldxa [%g0]ASI_MMU, %g1 */ 0x83307030, /* srlx %g1, CTXSHIFT, %g1 */ 0x02c04004, /* brz,pn %g1, .+0x10 */ 0x03000000, /* sethi %hi(new_tpc), %g1 */ 0x82106000, /* or %g1, %lo(new_tpc), %g1 */ 0x30800002, /* ba,a .+0x8 */ 0x82106000, /* or %g1, %lo(new_tpc), %g1 */ 0x81904000, /* wrpr %g1, %g0, %tpc */ 0x82006004, /* add %g1, 4, %g1 */ 0x83904000, /* wrpr %g1, %g0, %tnpc */ 0x03000000, /* sethi %hi(tmptick), %g1 */ 0x85410000, /* rd %tick, %g2 */ 0xc4706000, /* stx %g2, [%g1 + %lo(tmptick)] */ 0x30800000, /* ba,a addr */ NOP, NOP, NOP #else /* sun4v */ 0x82102008, /* mov SCRATCHPAD_CPUID, %g1 */ 0xc8d84400, /* ldxa [%g1]ASI_SCRATCHPAD, %g4 */ 0x89293000, /* sllx %g4, TSTAT_DATA_SHIFT, %g4 */ 0x03000000, /* sethi %hi(stat), %g1 */ 0x82004004, /* add %g1, %g4, %g1 */ 0xc4586000, /* ldx [%g1 + %lo(stat)], %g2 */ 0x8400a001, /* add %g2, 1, %g2 */ 0xc4706000, /* stx %g2, [%g1 + %lo(stat)] */ 0x85524000, /* rdpr %cwp, %g2 */ 0x87518000, /* rdpr %pstate, %g3 */ 0x8728f008, /* sllx %g3, 8, %g3 */ 0x84108003, /* or %g2, %g3, %g2 */ 0x8740c000, /* rd %asi, %g3 */ 0x8728f018, /* sllx %g3, 24, %g3 */ 0x83540000, /* rdpr %gl, %g1 */ 0x83287028, /* sllx %g1, 40, %g1 */ 0x86104003, /* or %g1, %g3, %g3 */ 0x84108003, /* or %g2, %g3, %g2 */ 0x8350c000, /* rdpr %tt, %g1 */ 0x8f902002, /* wrpr %g0, 2, %tl */ 0x85908000, /* wrpr %g2, %g0, %tstate */ 0x87904000, /* wrpr %g1, %g0, %tt */ 0xc2d80400, /* ldxa [%g0]ASI_SCRATCHPAD, %g1 */ 0xc2586000, /* ldx [%g1 + MMFSA_?_CTX], %g1 */ 0x02c04004, /* brz,pn %g1, .+0x10 */ 0x03000000, /* sethi %hi(new_tpc), %g1 */ 0x82106000, /* or %g1, %lo(new_tpc), %g1 */ 0x30800002, /* ba,a .+0x8 */ 0x82106000, /* or %g1, %lo(new_tpc), %g1 */ 0x81904000, /* wrpr %g1, %g0, %tpc */ 0x82006004, /* add %g1, 4, %g1 */ 0x83904000, /* wrpr %g1, %g0, %tnpc */ 0x03000000, /* sethi %hi(tmptick), %g1 */ 0x82004004, /* add %g1, %g4, %g1 */ 0x85410000, /* rd %tick, %g2 */ 0xc4706000, /* stx %g2, [%g1 + %lo(tmptick)] */ 0x30800000 /* ba,a addr */ #endif /* sun4v */ }; ASSERT(MUTEX_HELD(&tstat_lock)); #ifndef sun4v ASSERT(entno == TSTAT_ENT_ITLBMISS || entno == TSTAT_ENT_DTLBMISS); stat = TSTAT_DATA_OFFS(tcpu, tdata_traps) + entoffs; tmptick = TSTAT_DATA_OFFS(tcpu, tdata_tmptick); #else /* sun4v */ ASSERT(entno == TSTAT_ENT_ITLBMISS || entno == TSTAT_ENT_DTLBMISS || entno == TSTAT_ENT_IMMUMISS || entno == TSTAT_ENT_DMMUMISS); stat = TSTAT_CPU0_DATA_OFFS(tcpu, tdata_traps) + entoffs; tmptick = TSTAT_CPU0_DATA_OFFS(tcpu, tdata_tmptick); #endif /* sun4v */ if (itlb) { ret = &tcpu->tcpu_instr->tinst_itlbret; udata = &data->tpgsz_user.tmode_itlb; kdata = &data->tpgsz_kernel.tmode_itlb; tpc = TSTAT_INSTR_OFFS(tcpu, tinst_itlbret.ttlbr_ktlb); } else { ret = &tcpu->tcpu_instr->tinst_dtlbret; udata = &data->tpgsz_user.tmode_dtlb; kdata = &data->tpgsz_kernel.tmode_dtlb; tpc = TSTAT_INSTR_OFFS(tcpu, tinst_dtlbret.ttlbr_ktlb); } utpc = tpc + offsetof(tstat_tlbret_t, ttlbr_utlb) - offsetof(tstat_tlbret_t, ttlbr_ktlb); ASSERT(HI22(tpc) == HI22(utpc)); ent = (uint32_t *)((uintptr_t)tcpu->tcpu_instr + entoffs); orig = KERNELBASE + entoffs; va = (uintptr_t)tcpu->tcpu_ibase + entoffs; baoffs = TSTAT_TLBENT_BA * sizeof (uint32_t); #ifdef sun4v /* * Because of lack of space, interposing tlbent trap handler * for TLB and MMU miss traps cannot be placed in-line. Instead, * we copy it to the space set aside for shared trap handlers * continuation in the interposing trap table and invoke it by * placing a branch in the trap table itself. */ tent = ent; /* trap vector entry */ tentva = va; /* trap vector entry va */ if (itlb) { ent = (uint32_t *)((uintptr_t) &tcpu->tcpu_instr->tinst_immumiss); va = TSTAT_INSTR_OFFS(tcpu, tinst_immumiss); } else { ent = (uint32_t *)((uintptr_t) &tcpu->tcpu_instr->tinst_dmmumiss); va = TSTAT_INSTR_OFFS(tcpu, tinst_dmmumiss); } bcopy(mmumiss, tent, sizeof (mmumiss)); tent[0] |= DISP22(tentva, va); #endif /* sun4v */ bcopy(tlbent, ent, sizeof (tlbent)); #if defined(sun4v) ent[TSTAT_TLBENT_TDATASHFT] |= LO10((uintptr_t)TSTAT_DATA_SHIFT); #endif ent[TSTAT_TLBENT_STATHI] |= HI22(stat); ent[TSTAT_TLBENT_STATLO_LD] |= LO10(stat); ent[TSTAT_TLBENT_STATLO_ST] |= LO10(stat); #ifndef sun4v ent[TSTAT_TLBENT_MMUASI] |= asi; #else ent[TSTAT_TLBENT_TAGTARGET] |= tagtarget_off; #endif ent[TSTAT_TLBENT_TPCHI] |= HI22(tpc); ent[TSTAT_TLBENT_TPCLO_USER] |= LO10(utpc); ent[TSTAT_TLBENT_TPCLO_KERN] |= LO10(tpc); ent[TSTAT_TLBENT_TSHI] |= HI22(tmptick); ent[TSTAT_TLBENT_TSLO] |= LO10(tmptick); ent[TSTAT_TLBENT_BA] |= DISP22(va + baoffs, orig); /* * And now set up the TLB return entries. */ trapstat_tlbretent(tcpu, &ret->ttlbr_ktlb, &kdata->ttlb_tlb); trapstat_tlbretent(tcpu, &ret->ttlbr_ktsb, &kdata->ttlb_tsb); trapstat_tlbretent(tcpu, &ret->ttlbr_utlb, &udata->ttlb_tlb); trapstat_tlbretent(tcpu, &ret->ttlbr_utsb, &udata->ttlb_tsb); } #if defined(sun4v) #undef TSTAT_TLBENT_TDATASHFT #endif #undef TSTAT_TLBENT_STATHI #undef TSTAT_TLBENT_STATLO_LD #undef TSTAT_TLBENT_STATLO_ST #ifndef sun4v #undef TSTAT_TLBENT_MMUASI #else #undef TSTAT_TLBENT_TAGTARGET #endif #undef TSTAT_TLBENT_TPCHI #undef TSTAT_TLBENT_TPCLO_USER #undef TSTAT_TLBENT_TPCLO_KERN #undef TSTAT_TLBENT_TSHI #undef TSTAT_TLBENT_TSLO #undef TSTAT_TLBENT_BA /* * The TSTAT_ENABLED_* constants define offsets in the enabled entry; the * TSTAT_DISABLED_BA constant defines an offset in the disabled entry. Both * sets of constants are used only in trapstat_make_traptab() (below) and * #undef'd immediately afterwards. Any change to "enabled" or "disabled" * in trapstat_make_traptab() will likely require changes to these constants. */ #ifndef sun4v #define TSTAT_ENABLED_STATHI 0 #define TSTAT_ENABLED_STATLO_LD 1 #define TSTAT_ENABLED_STATLO_ST 3 #define TSTAT_ENABLED_BA 4 #define TSTAT_DISABLED_BA 0 static void trapstat_make_traptab(tstat_percpu_t *tcpu) { uint32_t *ent; uint64_t *stat; uintptr_t orig, va, en_baoffs, dis_baoffs; int nent; /* * This is the entry in the interposing trap table for enabled trap * table entries. It loads a counter, increments it and stores it * back before branching to the actual trap table entry. */ static const uint32_t enabled[TSTAT_ENT_NINSTR] = { 0x03000000, /* sethi %hi(stat), %g1 */ 0xc4586000, /* ldx [%g1 + %lo(stat)], %g2 */ 0x8400a001, /* add %g2, 1, %g2 */ 0xc4706000, /* stx %g2, [%g1 + %lo(stat)] */ 0x30800000, /* ba,a addr */ NOP, NOP, NOP }; /* * This is the entry in the interposing trap table for disabled trap * table entries. It simply branches to the actual, underlying trap * table entry. As explained in the "Implementation Details" section * of the block comment, all TL>0 traps _must_ use the disabled entry; * additional entries may be explicitly disabled through the use * of TSTATIOC_ENTRY/TSTATIOC_NOENTRY. */ static const uint32_t disabled[TSTAT_ENT_NINSTR] = { 0x30800000, /* ba,a addr */ NOP, NOP, NOP, NOP, NOP, NOP, NOP, }; ASSERT(MUTEX_HELD(&tstat_lock)); ent = tcpu->tcpu_instr->tinst_traptab; stat = (uint64_t *)TSTAT_DATA_OFFS(tcpu, tdata_traps); orig = KERNELBASE; va = (uintptr_t)tcpu->tcpu_ibase; en_baoffs = TSTAT_ENABLED_BA * sizeof (uint32_t); dis_baoffs = TSTAT_DISABLED_BA * sizeof (uint32_t); for (nent = 0; nent < TSTAT_TOTAL_NENT; nent++) { if (tstat_enabled[nent]) { bcopy(enabled, ent, sizeof (enabled)); ent[TSTAT_ENABLED_STATHI] |= HI22((uintptr_t)stat); ent[TSTAT_ENABLED_STATLO_LD] |= LO10((uintptr_t)stat); ent[TSTAT_ENABLED_STATLO_ST] |= LO10((uintptr_t)stat); ent[TSTAT_ENABLED_BA] |= DISP22(va + en_baoffs, orig); } else { bcopy(disabled, ent, sizeof (disabled)); ent[TSTAT_DISABLED_BA] |= DISP22(va + dis_baoffs, orig); } stat++; orig += sizeof (enabled); ent += sizeof (enabled) / sizeof (*ent); va += sizeof (enabled); } } #undef TSTAT_ENABLED_STATHI #undef TSTAT_ENABLED_STATLO_LD #undef TSTAT_ENABLED_STATLO_ST #undef TSTAT_ENABLED_BA #undef TSTAT_DISABLED_BA #else /* sun4v */ #define TSTAT_ENABLED_STATHI 0 #define TSTAT_ENABLED_STATLO 1 #define TSTAT_ENABLED_ADDRHI 2 #define TSTAT_ENABLED_ADDRLO 3 #define TSTAT_ENABLED_CONTBA 6 #define TSTAT_ENABLED_TDATASHFT 7 #define TSTAT_DISABLED_BA 0 static void trapstat_make_traptab(tstat_percpu_t *tcpu) { uint32_t *ent; uint64_t *stat; uintptr_t orig, va, en_baoffs, dis_baoffs; uintptr_t tstat_cont_va; int nent; /* * This is the entry in the interposing trap table for enabled trap * table entries. It loads a counter, increments it and stores it * back before branching to the actual trap table entry. * * All CPUs share the same interposing trap entry to count the * number of traps. Note that the trap counter is kept in per CPU * trap statistics area. Its address is obtained dynamically by * adding the offset of that CPU's trap statistics area from CPU 0 * (i.e. cpu_id * TSTAT_DATA_SIZE) to the address of the CPU 0 * trap counter already coded in the interposing trap entry itself. * * Since this interposing code sequence to count traps takes more * than 8 instructions, it's split in two parts as follows: * * tstat_trapcnt: * sethi %hi(stat), %g1 * or %g1, %lo[stat), %g1 ! %g1 = CPU0 trap counter addr * sethi %hi(addr), %g2 * or %g2, %lo(addr), %g2 ! %g2 = real trap handler addr * mov ASI_SCRATCHPAD_CPUID, %g3 * ldxa [%g3]ASI_SCRATCHPAD, %g3 ! %g3 = CPU ID * ba tstat_trapcnt_cont ! branch to tstat_trapcnt_cont * sllx %g3, TSTAT_DATA_SHIFT, %g3 ! %g3 = CPU trapstat data offset * * tstat_trapcnt_cont: * ldx [%g1 + %g3], %g4 ! get counter value * add %g4, 1, %g4 ! increment value * jmp %g2 ! jump to original trap handler * stx %g4, [%g1 + %g3] ! store counter value * * First part, i.e. tstat_trapcnt, is per trap and is kept in-line in * the interposing trap table. However, the tstat_trapcnt_cont code * sequence is shared by all traps and is kept right after the * the interposing trap table. */ static const uint32_t enabled[TSTAT_ENT_NINSTR] = { 0x03000000, /* sethi %hi(stat), %g1 */ 0x82106000, /* or %g1, %lo[stat), %g1 */ 0x05000000, /* sethi %hi(addr), %g2 */ 0x8410a000, /* or %g2, %lo(addr), %g2 */ 0x86102008, /* mov ASI_SCRATCHPAD_CPUID, %g3 */ 0xc6d8c400, /* ldxa [%g3]ASI_SCRATCHPAD, %g3 */ 0x10800000, /* ba enabled_cont */ 0x8728f000 /* sllx %g3, TSTAT_DATA_SHIFT, %g3 */ }; static const uint32_t enabled_cont[TSTAT_ENT_NINSTR] = { 0xc8584003, /* ldx [%g1 + %g3], %g4 */ 0x88012001, /* add %g4, 1, %g4 */ 0x81c08000, /* jmp %g2 */ 0xc8704003, /* stx %g4, [%g1 + %g3] */ NOP, NOP, NOP, NOP }; /* * This is the entry in the interposing trap table for disabled trap * table entries. It simply branches to the actual, underlying trap * table entry. As explained in the "Implementation Details" section * of the block comment, all TL>0 traps _must_ use the disabled entry; * additional entries may be explicitly disabled through the use * of TSTATIOC_ENTRY/TSTATIOC_NOENTRY. */ static const uint32_t disabled[TSTAT_ENT_NINSTR] = { 0x30800000, /* ba,a addr */ NOP, NOP, NOP, NOP, NOP, NOP, NOP, }; ASSERT(MUTEX_HELD(&tstat_lock)); ent = tcpu->tcpu_instr->tinst_traptab; stat = (uint64_t *)TSTAT_CPU0_DATA_OFFS(tcpu, tdata_traps); orig = KERNELBASE; va = (uintptr_t)tcpu->tcpu_ibase; en_baoffs = TSTAT_ENABLED_CONTBA * sizeof (uint32_t); dis_baoffs = TSTAT_DISABLED_BA * sizeof (uint32_t); tstat_cont_va = TSTAT_INSTR_OFFS(tcpu, tinst_trapcnt); for (nent = 0; nent < TSTAT_TOTAL_NENT; nent++) { if (tstat_enabled[nent]) { bcopy(enabled, ent, sizeof (enabled)); ent[TSTAT_ENABLED_STATHI] |= HI22((uintptr_t)stat); ent[TSTAT_ENABLED_STATLO] |= LO10((uintptr_t)stat); ent[TSTAT_ENABLED_ADDRHI] |= HI22((uintptr_t)orig); ent[TSTAT_ENABLED_ADDRLO] |= LO10((uintptr_t)orig); ent[TSTAT_ENABLED_CONTBA] |= DISP22(va + en_baoffs, tstat_cont_va); ent[TSTAT_ENABLED_TDATASHFT] |= LO10((uintptr_t)TSTAT_DATA_SHIFT); } else { bcopy(disabled, ent, sizeof (disabled)); ent[TSTAT_DISABLED_BA] |= DISP22(va + dis_baoffs, orig); } stat++; orig += sizeof (enabled); ent += sizeof (enabled) / sizeof (*ent); va += sizeof (enabled); } bcopy(enabled_cont, (uint32_t *)tcpu->tcpu_instr->tinst_trapcnt, sizeof (enabled_cont)); } #undef TSTAT_ENABLED_TDATASHFT #undef TSTAT_ENABLED_STATHI #undef TSTAT_ENABLED_STATLO #undef TSTAT_ENABLED_ADDRHI #undef TSTAT_ENABLED_ADDRLO #undef TSTAT_ENABLED_CONTBA #undef TSTAT_DISABLED_BA #endif /* sun4v */ #ifndef sun4v /* * See Section A.6 in SPARC v9 Manual. * max branch = 4*((2^21)-1) = 8388604 */ #define MAX_BICC_BRANCH_DISPLACEMENT (4 * ((1 << 21) - 1)) #endif static void trapstat_setup(processorid_t cpu) { tstat_percpu_t *tcpu = &tstat_percpu[cpu]; #ifndef sun4v int i; caddr_t va; pfn_t *pfn; cpu_t *cp; uint_t strand_idx; size_t tstat_offset; #endif ASSERT(tcpu->tcpu_pfn == NULL); ASSERT(tcpu->tcpu_instr == NULL); ASSERT(tcpu->tcpu_data == NULL); ASSERT(tcpu->tcpu_flags & TSTAT_CPU_SELECTED); ASSERT(!(tcpu->tcpu_flags & TSTAT_CPU_ALLOCATED)); ASSERT(MUTEX_HELD(&cpu_lock)); ASSERT(MUTEX_HELD(&tstat_lock)); #ifndef sun4v /* * The lower fifteen bits of the %tba are always read as zero; we must * align our instruction base address appropriately. */ tstat_offset = tstat_total_size; cp = cpu_get(cpu); ASSERT(cp != NULL); if ((strand_idx = cpu ^ pg_plat_hw_instance_id(cp, PGHW_IPIPE)) != 0) { /* * On sun4u platforms with multiple CPUs sharing the MMU * (Olympus-C has 2 strands per core), each CPU uses a * disjoint trap table. The indexing is based on the * strand id, which is obtained by XOR'ing the cpuid with * the coreid. */ tstat_offset += tstat_total_size * strand_idx; /* * Offset must be less than the maximum PC-relative branch * displacement for Bicc variants. See the Implementation * Details comment. */ ASSERT(tstat_offset <= MAX_BICC_BRANCH_DISPLACEMENT); } tcpu->tcpu_ibase = (caddr_t)((KERNELBASE - tstat_offset) & TSTAT_TBA_MASK); tcpu->tcpu_dbase = tcpu->tcpu_ibase + TSTAT_INSTR_SIZE; tcpu->tcpu_vabase = tcpu->tcpu_ibase; tcpu->tcpu_pfn = vmem_alloc(tstat_arena, tstat_total_pages, VM_SLEEP); bzero(tcpu->tcpu_pfn, tstat_total_pages); pfn = tcpu->tcpu_pfn; tcpu->tcpu_instr = vmem_alloc(tstat_arena, TSTAT_INSTR_SIZE, VM_SLEEP); va = (caddr_t)tcpu->tcpu_instr; for (i = 0; i < TSTAT_INSTR_PAGES; i++, va += MMU_PAGESIZE) *pfn++ = hat_getpfnum(kas.a_hat, va); /* * We must be sure that the pages that we will use to examine the data * have the same virtual color as the pages to which the data is being * recorded, hence the alignment and phase constraints on the * allocation. */ tcpu->tcpu_data = vmem_xalloc(tstat_arena, tstat_data_size, shm_alignment, (uintptr_t)tcpu->tcpu_dbase & (shm_alignment - 1), 0, 0, NULL, VM_SLEEP); bzero(tcpu->tcpu_data, tstat_data_size); tcpu->tcpu_data->tdata_cpuid = cpu; va = (caddr_t)tcpu->tcpu_data; for (i = 0; i < tstat_data_pages; i++, va += MMU_PAGESIZE) *pfn++ = hat_getpfnum(kas.a_hat, va); /* * Now that we have all of the instruction and data pages allocated, * make the trap table from scratch. */ trapstat_make_traptab(tcpu); if (tstat_options & TSTAT_OPT_TLBDATA) { /* * TLB Statistics have been specified; set up the I- and D-TLB * entries and corresponding TLB return entries. */ trapstat_tlbent(tcpu, TSTAT_ENT_ITLBMISS); trapstat_tlbent(tcpu, TSTAT_ENT_DTLBMISS); } #else /* sun4v */ /* * The lower fifteen bits of the %tba are always read as zero; hence * it must be aligned at least on 512K boundary. */ tcpu->tcpu_vabase = (caddr_t)(KERNELBASE - MMU_PAGESIZE4M); tcpu->tcpu_ibase = tcpu->tcpu_vabase; tcpu->tcpu_dbase = tcpu->tcpu_ibase + TSTAT_INSTR_SIZE + cpu * TSTAT_DATA_SIZE; tcpu->tcpu_pfn = &tstat_pfn; tcpu->tcpu_instr = (tstat_instr_t *)tstat_va; tcpu->tcpu_data = (tstat_data_t *)(tstat_va + TSTAT_INSTR_SIZE + cpu * TSTAT_DATA_SIZE); bzero(tcpu->tcpu_data, TSTAT_DATA_SIZE); tcpu->tcpu_data->tdata_cpuid = cpu; /* * Now that we have all of the instruction and data pages allocated, * make the trap table from scratch. It should be done only once * as it is shared by all CPUs. */ if (!tstat_traptab_initialized) trapstat_make_traptab(tcpu); if (tstat_options & TSTAT_OPT_TLBDATA) { /* * TLB Statistics have been specified; set up the I- and D-TLB * entries and corresponding TLB return entries. */ if (!tstat_traptab_initialized) { if (tstat_fast_tlbstat) { trapstat_tlbent(tcpu, TSTAT_ENT_IMMUMISS); trapstat_tlbent(tcpu, TSTAT_ENT_DMMUMISS); } else { trapstat_tlbent(tcpu, TSTAT_ENT_ITLBMISS); trapstat_tlbent(tcpu, TSTAT_ENT_DTLBMISS); } } } tstat_traptab_initialized = 1; #endif /* sun4v */ tcpu->tcpu_flags |= TSTAT_CPU_ALLOCATED; /* * Finally, get the target CPU to load the locked pages into its TLBs. */ xc_one(cpu, (xcfunc_t *)trapstat_load_tlb, 0, 0); } static void trapstat_teardown(processorid_t cpu) { tstat_percpu_t *tcpu = &tstat_percpu[cpu]; #ifndef sun4v int i; #endif caddr_t va = tcpu->tcpu_vabase; ASSERT(tcpu->tcpu_pfn != NULL); ASSERT(tcpu->tcpu_instr != NULL); ASSERT(tcpu->tcpu_data != NULL); ASSERT(tcpu->tcpu_flags & TSTAT_CPU_SELECTED); ASSERT(tcpu->tcpu_flags & TSTAT_CPU_ALLOCATED); ASSERT(!(tcpu->tcpu_flags & TSTAT_CPU_ENABLED)); ASSERT(MUTEX_HELD(&cpu_lock)); ASSERT(MUTEX_HELD(&tstat_lock)); #ifndef sun4v vmem_free(tstat_arena, tcpu->tcpu_pfn, tstat_total_pages); vmem_free(tstat_arena, tcpu->tcpu_instr, TSTAT_INSTR_SIZE); vmem_free(tstat_arena, tcpu->tcpu_data, tstat_data_size); for (i = 0; i < tstat_total_pages; i++, va += MMU_PAGESIZE) { xt_one(cpu, vtag_flushpage_tl1, (uint64_t)va, (uint64_t)ksfmmup); } #else xt_one(cpu, vtag_unmap_perm_tl1, (uint64_t)va, KCONTEXT); #endif tcpu->tcpu_pfn = NULL; tcpu->tcpu_instr = NULL; tcpu->tcpu_data = NULL; tcpu->tcpu_flags &= ~TSTAT_CPU_ALLOCATED; } static int trapstat_go() { cpu_t *cp; mutex_enter(&cpu_lock); mutex_enter(&tstat_lock); if (tstat_running) { mutex_exit(&tstat_lock); mutex_exit(&cpu_lock); return (EBUSY); } #ifdef sun4v /* * Allocate large page to hold interposing tables. */ tstat_va = contig_mem_alloc(MMU_PAGESIZE4M); tstat_pfn = va_to_pfn(tstat_va); if (tstat_pfn == PFN_INVALID) { mutex_exit(&tstat_lock); mutex_exit(&cpu_lock); return (EAGAIN); } /* * For detailed TLB statistics, invoke CPU specific interface * to see if it supports a low overhead interface to collect * TSB hit statistics. If so, make set tstat_fast_tlbstat flag * to reflect that. */ if (tstat_options & TSTAT_OPT_TLBDATA) { int error; tstat_fast_tlbstat = B_FALSE; error = cpu_trapstat_conf(CPU_TSTATCONF_INIT); if (error == 0) tstat_fast_tlbstat = B_TRUE; else if (error != ENOTSUP) { contig_mem_free(tstat_va, MMU_PAGESIZE4M); mutex_exit(&tstat_lock); mutex_exit(&cpu_lock); return (error); } } #endif /* sun4v */ /* * First, perform any necessary hot patching. */ trapstat_hotpatch(); /* * Allocate the resources we'll need to measure probe effect. */ trapstat_probe_alloc(); cp = cpu_list; do { if (!(tstat_percpu[cp->cpu_id].tcpu_flags & TSTAT_CPU_SELECTED)) continue; trapstat_setup(cp->cpu_id); /* * Note that due to trapstat_probe()'s use of global data, * we determine the probe effect on each CPU serially instead * of in parallel with an xc_all(). */ xc_one(cp->cpu_id, (xcfunc_t *)trapstat_probe, 0, 0); } while ((cp = cp->cpu_next) != cpu_list); xc_all((xcfunc_t *)trapstat_enable, 0, 0); trapstat_probe_free(); tstat_running = 1; mutex_exit(&tstat_lock); mutex_exit(&cpu_lock); return (0); } static int trapstat_stop() { int i; mutex_enter(&cpu_lock); mutex_enter(&tstat_lock); if (!tstat_running) { mutex_exit(&tstat_lock); mutex_exit(&cpu_lock); return (ENXIO); } xc_all((xcfunc_t *)trapstat_disable, 0, 0); for (i = 0; i <= max_cpuid; i++) { if (tstat_percpu[i].tcpu_flags & TSTAT_CPU_ALLOCATED) trapstat_teardown(i); } #ifdef sun4v tstat_traptab_initialized = 0; if (tstat_options & TSTAT_OPT_TLBDATA) cpu_trapstat_conf(CPU_TSTATCONF_FINI); contig_mem_free(tstat_va, MMU_PAGESIZE4M); #endif trapstat_hotpatch(); tstat_running = 0; mutex_exit(&tstat_lock); mutex_exit(&cpu_lock); return (0); } /* * This is trapstat's DR CPU configuration callback. It's called (with * cpu_lock held) to unconfigure a newly powered-off CPU, or to configure a * powered-off CPU that is to be brought into the system. We need only take * action in the unconfigure case: because a powered-off CPU will have its * trap table restored to KERNELBASE if it is ever powered back on, we must * update the flags to reflect that trapstat is no longer enabled on the * powered-off CPU. Note that this means that a TSTAT_CPU_ENABLED CPU that * is unconfigured/powered off and later powered back on/reconfigured will * _not_ be re-TSTAT_CPU_ENABLED. */ static int trapstat_cpu_setup(cpu_setup_t what, processorid_t cpu) { tstat_percpu_t *tcpu = &tstat_percpu[cpu]; ASSERT(MUTEX_HELD(&cpu_lock)); mutex_enter(&tstat_lock); if (!tstat_running) { mutex_exit(&tstat_lock); return (0); } switch (what) { case CPU_CONFIG: ASSERT(!(tcpu->tcpu_flags & TSTAT_CPU_ENABLED)); break; case CPU_UNCONFIG: if (tcpu->tcpu_flags & TSTAT_CPU_ENABLED) { tcpu->tcpu_flags &= ~TSTAT_CPU_ENABLED; #ifdef sun4v /* * A power-off, causes the cpu mondo queues to be * unconfigured on sun4v. Since we can't teardown * trapstat's mappings on the cpu that is going away, * we simply mark it as not allocated. This will * prevent a teardown on a cpu with the same cpu id * that might have been added while trapstat is running. */ if (tcpu->tcpu_flags & TSTAT_CPU_ALLOCATED) { tcpu->tcpu_pfn = NULL; tcpu->tcpu_instr = NULL; tcpu->tcpu_data = NULL; tcpu->tcpu_flags &= ~TSTAT_CPU_ALLOCATED; } #endif } break; default: break; } mutex_exit(&tstat_lock); return (0); } /* * This is called before a CPR suspend and after a CPR resume. We don't have * anything to do before a suspend, but after a restart we must restore the * trap table to be our interposing trap table. However, we don't actually * know whether or not the CPUs have been powered off -- this routine may be * called while restoring from a failed CPR suspend. We thus run through each * TSTAT_CPU_ENABLED CPU, and explicitly destroy and reestablish its * interposing trap table. This assures that our state is correct regardless * of whether or not the CPU has been newly powered on. */ /*ARGSUSED*/ static boolean_t trapstat_cpr(void *arg, int code) { cpu_t *cp; if (code == CB_CODE_CPR_CHKPT) return (B_TRUE); ASSERT(code == CB_CODE_CPR_RESUME); mutex_enter(&cpu_lock); mutex_enter(&tstat_lock); if (!tstat_running) { mutex_exit(&tstat_lock); mutex_exit(&cpu_lock); return (B_TRUE); } cp = cpu_list; do { tstat_percpu_t *tcpu = &tstat_percpu[cp->cpu_id]; if (!(tcpu->tcpu_flags & TSTAT_CPU_ENABLED)) continue; ASSERT(tcpu->tcpu_flags & TSTAT_CPU_SELECTED); ASSERT(tcpu->tcpu_flags & TSTAT_CPU_ALLOCATED); xc_one(cp->cpu_id, (xcfunc_t *)trapstat_disable, 0, 0); ASSERT(!(tcpu->tcpu_flags & TSTAT_CPU_ENABLED)); /* * Preserve this CPU's data in tstat_buffer and rip down its * interposing trap table. */ bcopy(tcpu->tcpu_data, tstat_buffer, tstat_data_t_size); trapstat_teardown(cp->cpu_id); ASSERT(!(tcpu->tcpu_flags & TSTAT_CPU_ALLOCATED)); /* * Reestablish the interposing trap table and restore the old * data. */ trapstat_setup(cp->cpu_id); ASSERT(tcpu->tcpu_flags & TSTAT_CPU_ALLOCATED); bcopy(tstat_buffer, tcpu->tcpu_data, tstat_data_t_size); xc_one(cp->cpu_id, (xcfunc_t *)trapstat_enable, 0, 0); } while ((cp = cp->cpu_next) != cpu_list); mutex_exit(&tstat_lock); mutex_exit(&cpu_lock); return (B_TRUE); } /*ARGSUSED*/ static int trapstat_open(dev_t *devp, int flag, int otyp, cred_t *cred_p) { int i; mutex_enter(&cpu_lock); mutex_enter(&tstat_lock); if (tstat_open != 0) { mutex_exit(&tstat_lock); mutex_exit(&cpu_lock); return (EBUSY); } /* * Register this in open() rather than in attach() to prevent deadlock * with DR code. During attach, I/O device tree locks are grabbed * before trapstat_attach() is invoked - registering in attach * will result in the lock order: device tree lock, cpu_lock. * DR code however requires that cpu_lock be acquired before * device tree locks. */ ASSERT(!tstat_running); register_cpu_setup_func((cpu_setup_func_t *)trapstat_cpu_setup, NULL); /* * Clear all options. And until specific CPUs are specified, we'll * mark all CPUs as selected. */ tstat_options = 0; for (i = 0; i <= max_cpuid; i++) tstat_percpu[i].tcpu_flags |= TSTAT_CPU_SELECTED; /* * By default, all traps at TL=0 are enabled. Traps at TL>0 must * be disabled. */ for (i = 0; i < TSTAT_TOTAL_NENT; i++) tstat_enabled[i] = i < TSTAT_NENT ? 1 : 0; tstat_open = 1; mutex_exit(&tstat_lock); mutex_exit(&cpu_lock); return (0); } /*ARGSUSED*/ static int trapstat_close(dev_t dev, int flag, int otyp, cred_t *cred_p) { (void) trapstat_stop(); ASSERT(!tstat_running); mutex_enter(&cpu_lock); unregister_cpu_setup_func((cpu_setup_func_t *)trapstat_cpu_setup, NULL); mutex_exit(&cpu_lock); tstat_open = 0; return (DDI_SUCCESS); } static int trapstat_option(int option) { mutex_enter(&tstat_lock); if (tstat_running) { mutex_exit(&tstat_lock); return (EBUSY); } tstat_options |= option; mutex_exit(&tstat_lock); return (0); } /*ARGSUSED*/ static int trapstat_ioctl(dev_t dev, int cmd, intptr_t arg, int md, cred_t *crd, int *rval) { int i, j, out; size_t dsize; switch (cmd) { case TSTATIOC_GO: return (trapstat_go()); case TSTATIOC_NOGO: return (trapstat_option(TSTAT_OPT_NOGO)); case TSTATIOC_STOP: return (trapstat_stop()); case TSTATIOC_CPU: if (arg < 0 || arg > max_cpuid) return (EINVAL); /*FALLTHROUGH*/ case TSTATIOC_NOCPU: mutex_enter(&tstat_lock); if (tstat_running) { mutex_exit(&tstat_lock); return (EBUSY); } /* * If this is the first CPU to be specified (or if we are * being asked to explicitly de-select CPUs), disable all CPUs. */ if (!(tstat_options & TSTAT_OPT_CPU) || cmd == TSTATIOC_NOCPU) { tstat_options |= TSTAT_OPT_CPU; for (i = 0; i <= max_cpuid; i++) { tstat_percpu_t *tcpu = &tstat_percpu[i]; ASSERT(cmd == TSTATIOC_NOCPU || (tcpu->tcpu_flags & TSTAT_CPU_SELECTED)); tcpu->tcpu_flags &= ~TSTAT_CPU_SELECTED; } } if (cmd == TSTATIOC_CPU) tstat_percpu[arg].tcpu_flags |= TSTAT_CPU_SELECTED; mutex_exit(&tstat_lock); return (0); case TSTATIOC_ENTRY: mutex_enter(&tstat_lock); if (tstat_running) { mutex_exit(&tstat_lock); return (EBUSY); } if (arg >= TSTAT_NENT || arg < 0) { mutex_exit(&tstat_lock); return (EINVAL); } if (!(tstat_options & TSTAT_OPT_ENTRY)) { /* * If this is the first entry that we are explicitly * enabling, explicitly disable every TL=0 entry. */ for (i = 0; i < TSTAT_NENT; i++) tstat_enabled[i] = 0; tstat_options |= TSTAT_OPT_ENTRY; } tstat_enabled[arg] = 1; mutex_exit(&tstat_lock); return (0); case TSTATIOC_NOENTRY: mutex_enter(&tstat_lock); if (tstat_running) { mutex_exit(&tstat_lock); return (EBUSY); } for (i = 0; i < TSTAT_NENT; i++) tstat_enabled[i] = 0; mutex_exit(&tstat_lock); return (0); case TSTATIOC_READ: mutex_enter(&tstat_lock); if (tstat_options & TSTAT_OPT_TLBDATA) { dsize = tstat_data_t_exported_size; } else { dsize = sizeof (tstat_data_t); } for (i = 0, out = 0; i <= max_cpuid; i++) { tstat_percpu_t *tcpu = &tstat_percpu[i]; if (!(tcpu->tcpu_flags & TSTAT_CPU_ENABLED)) continue; ASSERT(tcpu->tcpu_flags & TSTAT_CPU_SELECTED); ASSERT(tcpu->tcpu_flags & TSTAT_CPU_ALLOCATED); tstat_buffer->tdata_cpuid = -1; xc_one(i, (xcfunc_t *)trapstat_snapshot, 0, 0); if (tstat_buffer->tdata_cpuid == -1) { /* * This CPU is not currently responding to * cross calls; we have caught it while it is * being unconfigured. We'll drop tstat_lock * and pick up and drop cpu_lock. By the * time we acquire cpu_lock, the DR operation * will appear consistent and we can assert * that trapstat_cpu_setup() has cleared * TSTAT_CPU_ENABLED. */ mutex_exit(&tstat_lock); mutex_enter(&cpu_lock); mutex_exit(&cpu_lock); mutex_enter(&tstat_lock); ASSERT(!(tcpu->tcpu_flags & TSTAT_CPU_ENABLED)); continue; } /* * Need to compensate for the difference between page * sizes exported to users and page sizes available * within the kernel. */ if ((tstat_options & TSTAT_OPT_TLBDATA) && (tstat_pgszs != tstat_user_pgszs)) { tstat_pgszdata_t *tp; uint_t szc; tp = &tstat_buffer->tdata_pgsz[0]; for (j = 0; j < tstat_user_pgszs; j++) { if ((szc = USERSZC_2_SZC(j)) != j) { bcopy(&tp[szc], &tp[j], sizeof (tstat_pgszdata_t)); } } } if (copyout(tstat_buffer, (void *)arg, dsize) != 0) { mutex_exit(&tstat_lock); return (EFAULT); } out++; arg += dsize; } if (out != max_cpuid + 1) { processorid_t cpuid = -1; arg += offsetof(tstat_data_t, tdata_cpuid); if (copyout(&cpuid, (void *)arg, sizeof (cpuid)) != 0) { mutex_exit(&tstat_lock); return (EFAULT); } } mutex_exit(&tstat_lock); return (0); case TSTATIOC_TLBDATA: return (trapstat_option(TSTAT_OPT_TLBDATA)); default: break; } return (ENOTTY); } /*ARGSUSED*/ static int trapstat_info(dev_info_t *dip, ddi_info_cmd_t infocmd, void *arg, void **result) { int error; switch (infocmd) { case DDI_INFO_DEVT2DEVINFO: *result = (void *)tstat_devi; error = DDI_SUCCESS; break; case DDI_INFO_DEVT2INSTANCE: *result = (void *)0; error = DDI_SUCCESS; break; default: error = DDI_FAILURE; } return (error); } static int trapstat_attach(dev_info_t *devi, ddi_attach_cmd_t cmd) { switch (cmd) { case DDI_ATTACH: break; case DDI_RESUME: return (DDI_SUCCESS); default: return (DDI_FAILURE); } if (ddi_create_minor_node(devi, "trapstat", S_IFCHR, 0, DDI_PSEUDO, 0) == DDI_FAILURE) { ddi_remove_minor_node(devi, NULL); return (DDI_FAILURE); } ddi_report_dev(devi); tstat_devi = devi; tstat_pgszs = page_num_pagesizes(); tstat_user_pgszs = page_num_user_pagesizes(0); tstat_data_t_size = sizeof (tstat_data_t) + (tstat_pgszs - 1) * sizeof (tstat_pgszdata_t); tstat_data_t_exported_size = sizeof (tstat_data_t) + (tstat_user_pgszs - 1) * sizeof (tstat_pgszdata_t); #ifndef sun4v tstat_data_pages = (tstat_data_t_size >> MMU_PAGESHIFT) + 1; tstat_total_pages = TSTAT_INSTR_PAGES + tstat_data_pages; tstat_data_size = tstat_data_pages * MMU_PAGESIZE; tstat_total_size = TSTAT_INSTR_SIZE + tstat_data_size; #else ASSERT(tstat_data_t_size <= TSTAT_DATA_SIZE); #endif tstat_percpu = kmem_zalloc((max_cpuid + 1) * sizeof (tstat_percpu_t), KM_SLEEP); /* * Create our own arena backed by segkmem to assure a source of * MMU_PAGESIZE-aligned allocations. We allocate out of the * heap32_arena to assure that we can address the allocated memory with * a single sethi/simm13 pair in the interposing trap table entries. */ tstat_arena = vmem_create("trapstat", NULL, 0, MMU_PAGESIZE, segkmem_alloc_permanent, segkmem_free, heap32_arena, 0, VM_SLEEP); tstat_enabled = kmem_alloc(TSTAT_TOTAL_NENT * sizeof (int), KM_SLEEP); tstat_buffer = kmem_alloc(tstat_data_t_size, KM_SLEEP); /* * CB_CL_CPR_POST_USER is the class that executes from cpr_resume() * after user threads can be restarted. By executing in this class, * we are assured of the availability of system services needed to * resume trapstat (specifically, we are assured that all CPUs are * restarted and responding to cross calls). */ tstat_cprcb = callb_add(trapstat_cpr, NULL, CB_CL_CPR_POST_USER, "trapstat"); return (DDI_SUCCESS); } static int trapstat_detach(dev_info_t *devi, ddi_detach_cmd_t cmd) { int rval; ASSERT(devi == tstat_devi); switch (cmd) { case DDI_DETACH: break; case DDI_SUSPEND: return (DDI_SUCCESS); default: return (DDI_FAILURE); } ASSERT(!tstat_running); rval = callb_delete(tstat_cprcb); ASSERT(rval == 0); kmem_free(tstat_buffer, tstat_data_t_size); kmem_free(tstat_enabled, TSTAT_TOTAL_NENT * sizeof (int)); vmem_destroy(tstat_arena); kmem_free(tstat_percpu, (max_cpuid + 1) * sizeof (tstat_percpu_t)); ddi_remove_minor_node(devi, NULL); return (DDI_SUCCESS); } /* * Configuration data structures */ static struct cb_ops trapstat_cb_ops = { trapstat_open, /* open */ trapstat_close, /* close */ nulldev, /* strategy */ nulldev, /* print */ nodev, /* dump */ nodev, /* read */ nodev, /* write */ trapstat_ioctl, /* ioctl */ nodev, /* devmap */ nodev, /* mmap */ nodev, /* segmap */ nochpoll, /* poll */ ddi_prop_op, /* cb_prop_op */ 0, /* streamtab */ D_MP | D_NEW /* Driver compatibility flag */ }; static struct dev_ops trapstat_ops = { DEVO_REV, /* devo_rev, */ 0, /* refcnt */ trapstat_info, /* getinfo */ nulldev, /* identify */ nulldev, /* probe */ trapstat_attach, /* attach */ trapstat_detach, /* detach */ nulldev, /* reset */ &trapstat_cb_ops, /* cb_ops */ (struct bus_ops *)0, /* bus_ops */ NULL, /* power */ ddi_quiesce_not_needed, /* quiesce */ }; static struct modldrv modldrv = { &mod_driverops, /* Type of module. This one is a driver */ "Trap Statistics", /* name of module */ &trapstat_ops, /* driver ops */ }; static struct modlinkage modlinkage = { MODREV_1, (void *)&modldrv, NULL }; int _init(void) { return (mod_install(&modlinkage)); } int _fini(void) { return (mod_remove(&modlinkage)); } int _info(struct modinfo *modinfop) { return (mod_info(&modlinkage, modinfop)); }