These XIVE tests include: - General interrupt IRQ tests that: - enable and trigger an interrupt - acknowledge the interrupt - end of interrupt processing - Test the Pull Thread Context to Odd Thread Reporting Line - Test the different cache flush inject and queue sync inject operations Co-authored-by: Frederic Barrat <fbarrat@linux.ibm.com> Co-authored-by: Glenn Miles <milesg@linux.ibm.com> Co-authored-by: Michael Kowal <kowal@linux.ibm.com> Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com> Signed-off-by: Glenn Miles <milesg@linux.ibm.com> Signed-off-by: Michael Kowal <kowal@linux.ibm.com> Reviewed-by: Cédric Le Goater <clg@redhat.com> Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
		
			
				
	
	
		
			345 lines
		
	
	
		
			10 KiB
		
	
	
	
		
			C
		
	
	
	
	
	
			
		
		
	
	
			345 lines
		
	
	
		
			10 KiB
		
	
	
	
		
			C
		
	
	
	
	
	
/*
 | 
						|
 * QTest testcase for PowerNV 10 interrupt controller (xive2)
 | 
						|
 *  - Test irq to hardware thread
 | 
						|
 *  - Test 'Pull Thread Context to Odd Thread Reporting Line'
 | 
						|
 *
 | 
						|
 * Copyright (c) 2024, IBM Corporation.
 | 
						|
 *
 | 
						|
 * SPDX-License-Identifier: GPL-2.0-or-later
 | 
						|
 */
 | 
						|
#include "qemu/osdep.h"
 | 
						|
#include "libqtest.h"
 | 
						|
 | 
						|
#include "pnv-xive2-common.h"
 | 
						|
#include "hw/intc/pnv_xive2_regs.h"
 | 
						|
#include "hw/ppc/xive_regs.h"
 | 
						|
#include "hw/ppc/xive2_regs.h"
 | 
						|
 | 
						|
#define SMT                     4 /* some tests will break if less than 4 */
 | 
						|
 | 
						|
 | 
						|
static void set_table(QTestState *qts, uint64_t type, uint64_t addr)
 | 
						|
{
 | 
						|
    uint64_t vsd, size, log_size;
 | 
						|
 | 
						|
    /*
 | 
						|
     * First, let's make sure that all the resources used fit in the
 | 
						|
     * given table.
 | 
						|
     */
 | 
						|
    switch (type) {
 | 
						|
    case VST_ESB:
 | 
						|
        size = MAX_IRQS / 4;
 | 
						|
        break;
 | 
						|
    case VST_EAS:
 | 
						|
        size = MAX_IRQS * 8;
 | 
						|
        break;
 | 
						|
    case VST_END:
 | 
						|
        size = MAX_ENDS * 32;
 | 
						|
        break;
 | 
						|
    case VST_NVP:
 | 
						|
    case VST_NVG:
 | 
						|
    case VST_NVC:
 | 
						|
        size = MAX_VPS * 32;
 | 
						|
        break;
 | 
						|
    case VST_SYNC:
 | 
						|
        size = 64 * 1024;
 | 
						|
        break;
 | 
						|
    default:
 | 
						|
        g_assert_not_reached();
 | 
						|
    }
 | 
						|
 | 
						|
    g_assert_cmpuint(size, <=, XIVE_VST_SIZE);
 | 
						|
    log_size = ctzl(XIVE_VST_SIZE) - 12;
 | 
						|
 | 
						|
    vsd = ((uint64_t) VSD_MODE_EXCLUSIVE) << 62 | addr | log_size;
 | 
						|
    pnv_xive_xscom_write(qts, X_VC_VSD_TABLE_ADDR, type << 48);
 | 
						|
    pnv_xive_xscom_write(qts, X_VC_VSD_TABLE_DATA, vsd);
 | 
						|
 | 
						|
    if (type != VST_EAS && type != VST_IC && type != VST_ERQ) {
 | 
						|
        pnv_xive_xscom_write(qts, X_PC_VSD_TABLE_ADDR, type << 48);
 | 
						|
        pnv_xive_xscom_write(qts, X_PC_VSD_TABLE_DATA, vsd);
 | 
						|
    }
 | 
						|
}
 | 
						|
 | 
						|
static void set_tima8(QTestState *qts, uint32_t pir, uint32_t offset,
 | 
						|
                      uint8_t b)
 | 
						|
{
 | 
						|
    uint64_t ic_addr;
 | 
						|
 | 
						|
    ic_addr = XIVE_IC_TM_INDIRECT + (pir << XIVE_PAGE_SHIFT);
 | 
						|
    qtest_writeb(qts, ic_addr + offset, b);
 | 
						|
}
 | 
						|
 | 
						|
static void set_tima32(QTestState *qts, uint32_t pir, uint32_t offset,
 | 
						|
                       uint32_t l)
 | 
						|
{
 | 
						|
    uint64_t ic_addr;
 | 
						|
 | 
						|
    ic_addr = XIVE_IC_TM_INDIRECT + (pir << XIVE_PAGE_SHIFT);
 | 
						|
    qtest_writel(qts, ic_addr + offset, l);
 | 
						|
}
 | 
						|
 | 
						|
static uint8_t get_tima8(QTestState *qts, uint32_t pir, uint32_t offset)
 | 
						|
{
 | 
						|
    uint64_t ic_addr;
 | 
						|
 | 
						|
    ic_addr = XIVE_IC_TM_INDIRECT + (pir << XIVE_PAGE_SHIFT);
 | 
						|
    return qtest_readb(qts, ic_addr + offset);
 | 
						|
}
 | 
						|
 | 
						|
static uint16_t get_tima16(QTestState *qts, uint32_t pir, uint32_t offset)
 | 
						|
{
 | 
						|
    uint64_t ic_addr;
 | 
						|
 | 
						|
    ic_addr = XIVE_IC_TM_INDIRECT + (pir << XIVE_PAGE_SHIFT);
 | 
						|
    return qtest_readw(qts, ic_addr + offset);
 | 
						|
}
 | 
						|
 | 
						|
static uint32_t get_tima32(QTestState *qts, uint32_t pir, uint32_t offset)
 | 
						|
{
 | 
						|
    uint64_t ic_addr;
 | 
						|
 | 
						|
    ic_addr = XIVE_IC_TM_INDIRECT + (pir << XIVE_PAGE_SHIFT);
 | 
						|
    return qtest_readl(qts, ic_addr + offset);
 | 
						|
}
 | 
						|
 | 
						|
static void reset_pool_threads(QTestState *qts)
 | 
						|
{
 | 
						|
    uint8_t first_group = 0;
 | 
						|
    int i;
 | 
						|
 | 
						|
    for (i = 0; i < SMT; i++) {
 | 
						|
        uint32_t nvp_idx = 0x100 + i;
 | 
						|
        set_nvp(qts, nvp_idx, first_group);
 | 
						|
        set_tima32(qts, i, TM_QW2_HV_POOL + TM_WORD0, 0x000000ff);
 | 
						|
        set_tima32(qts, i, TM_QW2_HV_POOL + TM_WORD1, 0);
 | 
						|
        set_tima32(qts, i, TM_QW2_HV_POOL + TM_WORD2, TM_QW2W2_VP | nvp_idx);
 | 
						|
    }
 | 
						|
}
 | 
						|
 | 
						|
static void reset_hw_threads(QTestState *qts)
 | 
						|
{
 | 
						|
    uint8_t first_group = 0;
 | 
						|
    uint32_t w1 = 0x000000ff;
 | 
						|
    int i;
 | 
						|
 | 
						|
    if (SMT >= 4) {
 | 
						|
        /* define 2 groups of 2, part of a bigger group of size 4 */
 | 
						|
        set_nvg(qts, 0x80, 0x02);
 | 
						|
        set_nvg(qts, 0x82, 0x02);
 | 
						|
        set_nvg(qts, 0x81, 0);
 | 
						|
        first_group = 0x01;
 | 
						|
        w1 = 0x000300ff;
 | 
						|
    }
 | 
						|
 | 
						|
    for (i = 0; i < SMT; i++) {
 | 
						|
        set_nvp(qts, 0x80 + i, first_group);
 | 
						|
        set_tima32(qts, i, TM_QW3_HV_PHYS + TM_WORD0, 0x00ff00ff);
 | 
						|
        set_tima32(qts, i, TM_QW3_HV_PHYS + TM_WORD1, w1);
 | 
						|
        set_tima32(qts, i, TM_QW3_HV_PHYS + TM_WORD2, 0x80000000);
 | 
						|
    }
 | 
						|
}
 | 
						|
 | 
						|
static void reset_state(QTestState *qts)
 | 
						|
{
 | 
						|
    size_t mem_used = XIVE_MEM_END - XIVE_MEM_START;
 | 
						|
 | 
						|
    qtest_memset(qts, XIVE_MEM_START, 0, mem_used);
 | 
						|
    reset_hw_threads(qts);
 | 
						|
    reset_pool_threads(qts);
 | 
						|
}
 | 
						|
 | 
						|
static void init_xive(QTestState *qts)
 | 
						|
{
 | 
						|
    uint64_t val1, val2, range;
 | 
						|
 | 
						|
    /*
 | 
						|
     * We can take a few shortcuts here, as we know the default values
 | 
						|
     * used for xive initialization
 | 
						|
     */
 | 
						|
 | 
						|
    /*
 | 
						|
     * Set the BARs.
 | 
						|
     * We reuse the same values used by firmware to ease debug.
 | 
						|
     */
 | 
						|
    pnv_xive_xscom_write(qts, X_CQ_IC_BAR, XIVE_IC_BAR);
 | 
						|
    pnv_xive_xscom_write(qts, X_CQ_TM_BAR, XIVE_TM_BAR);
 | 
						|
 | 
						|
    /* ESB and NVPG use 2 pages per resource. The others only one page */
 | 
						|
    range = (MAX_IRQS << 17) >> 25;
 | 
						|
    val1 = XIVE_ESB_BAR | range;
 | 
						|
    pnv_xive_xscom_write(qts, X_CQ_ESB_BAR, val1);
 | 
						|
 | 
						|
    range = (MAX_ENDS << 16) >> 25;
 | 
						|
    val1 = XIVE_END_BAR | range;
 | 
						|
    pnv_xive_xscom_write(qts, X_CQ_END_BAR, val1);
 | 
						|
 | 
						|
    range = (MAX_VPS << 17) >> 25;
 | 
						|
    val1 = XIVE_NVPG_BAR | range;
 | 
						|
    pnv_xive_xscom_write(qts, X_CQ_NVPG_BAR, val1);
 | 
						|
 | 
						|
    range = (MAX_VPS << 16) >> 25;
 | 
						|
    val1 = XIVE_NVC_BAR | range;
 | 
						|
    pnv_xive_xscom_write(qts, X_CQ_NVC_BAR, val1);
 | 
						|
 | 
						|
    /*
 | 
						|
     * Enable hw threads.
 | 
						|
     * We check the value written. Useless with current
 | 
						|
     * implementation, but it validates the xscom read path and it's
 | 
						|
     * what the hardware procedure says
 | 
						|
     */
 | 
						|
    val1 = 0xF000000000000000ull; /* core 0, 4 threads */
 | 
						|
    pnv_xive_xscom_write(qts, X_TCTXT_EN0, val1);
 | 
						|
    val2 = pnv_xive_xscom_read(qts, X_TCTXT_EN0);
 | 
						|
    g_assert_cmphex(val1, ==, val2);
 | 
						|
 | 
						|
    /* Memory tables */
 | 
						|
    set_table(qts, VST_ESB, XIVE_ESB_MEM);
 | 
						|
    set_table(qts, VST_EAS, XIVE_EAS_MEM);
 | 
						|
    set_table(qts, VST_END, XIVE_END_MEM);
 | 
						|
    set_table(qts, VST_NVP, XIVE_NVP_MEM);
 | 
						|
    set_table(qts, VST_NVG, XIVE_NVG_MEM);
 | 
						|
    set_table(qts, VST_NVC, XIVE_NVC_MEM);
 | 
						|
    set_table(qts, VST_SYNC, XIVE_SYNC_MEM);
 | 
						|
 | 
						|
    reset_hw_threads(qts);
 | 
						|
    reset_pool_threads(qts);
 | 
						|
}
 | 
						|
 | 
						|
static void test_hw_irq(QTestState *qts)
 | 
						|
{
 | 
						|
    uint32_t irq = 2;
 | 
						|
    uint32_t irq_data = 0x600df00d;
 | 
						|
    uint32_t end_index = 5;
 | 
						|
    uint32_t target_pir = 1;
 | 
						|
    uint32_t target_nvp = 0x80 + target_pir;
 | 
						|
    uint8_t priority = 5;
 | 
						|
    uint32_t reg32;
 | 
						|
    uint16_t reg16;
 | 
						|
    uint8_t pq, nsr, cppr;
 | 
						|
 | 
						|
    printf("# ============================================================\n");
 | 
						|
    printf("# Testing irq %d to hardware thread %d\n", irq, target_pir);
 | 
						|
 | 
						|
    /* irq config */
 | 
						|
    set_eas(qts, irq, end_index, irq_data);
 | 
						|
    set_end(qts, end_index, target_nvp, priority, false /* group */);
 | 
						|
 | 
						|
    /* enable and trigger irq */
 | 
						|
    get_esb(qts, irq, XIVE_EOI_PAGE, XIVE_ESB_SET_PQ_00);
 | 
						|
    set_esb(qts, irq, XIVE_TRIGGER_PAGE, 0, 0);
 | 
						|
 | 
						|
    /* check irq is raised on cpu */
 | 
						|
    pq = get_esb(qts, irq, XIVE_EOI_PAGE, XIVE_ESB_GET);
 | 
						|
    g_assert_cmpuint(pq, ==, XIVE_ESB_PENDING);
 | 
						|
 | 
						|
    reg32 = get_tima32(qts, target_pir, TM_QW3_HV_PHYS + TM_WORD0);
 | 
						|
    nsr = reg32 >> 24;
 | 
						|
    cppr = (reg32 >> 16) & 0xFF;
 | 
						|
    g_assert_cmphex(nsr, ==, 0x80);
 | 
						|
    g_assert_cmphex(cppr, ==, 0xFF);
 | 
						|
 | 
						|
    /* ack the irq */
 | 
						|
    reg16 = get_tima16(qts, target_pir, TM_SPC_ACK_HV_REG);
 | 
						|
    nsr = reg16 >> 8;
 | 
						|
    cppr = reg16 & 0xFF;
 | 
						|
    g_assert_cmphex(nsr, ==, 0x80);
 | 
						|
    g_assert_cmphex(cppr, ==, priority);
 | 
						|
 | 
						|
    /* check irq data is what was configured */
 | 
						|
    reg32 = qtest_readl(qts, xive_get_queue_addr(end_index));
 | 
						|
    g_assert_cmphex((reg32 & 0x7fffffff), ==, (irq_data & 0x7fffffff));
 | 
						|
 | 
						|
    /* End Of Interrupt */
 | 
						|
    set_esb(qts, irq, XIVE_EOI_PAGE, XIVE_ESB_STORE_EOI, 0);
 | 
						|
    pq = get_esb(qts, irq, XIVE_EOI_PAGE, XIVE_ESB_GET);
 | 
						|
    g_assert_cmpuint(pq, ==, XIVE_ESB_RESET);
 | 
						|
 | 
						|
    /* reset CPPR */
 | 
						|
    set_tima8(qts, target_pir, TM_QW3_HV_PHYS + TM_CPPR, 0xFF);
 | 
						|
    reg32 = get_tima32(qts, target_pir, TM_QW3_HV_PHYS + TM_WORD0);
 | 
						|
    nsr = reg32 >> 24;
 | 
						|
    cppr = (reg32 >> 16) & 0xFF;
 | 
						|
    g_assert_cmphex(nsr, ==, 0x00);
 | 
						|
    g_assert_cmphex(cppr, ==, 0xFF);
 | 
						|
}
 | 
						|
 | 
						|
#define XIVE_ODD_CL 0x80
 | 
						|
static void test_pull_thread_ctx_to_odd_thread_cl(QTestState *qts)
 | 
						|
{
 | 
						|
    uint32_t target_pir = 1;
 | 
						|
    uint32_t target_nvp = 0x80 + target_pir;
 | 
						|
    Xive2Nvp nvp;
 | 
						|
    uint8_t cl_pair[XIVE_REPORT_SIZE];
 | 
						|
    uint32_t qw1w0, qw3w0, qw1w2, qw2w2;
 | 
						|
    uint8_t qw3b8;
 | 
						|
    uint32_t cl_word;
 | 
						|
    uint32_t word2;
 | 
						|
 | 
						|
    printf("# ============================================================\n");
 | 
						|
    printf("# Testing 'Pull Thread Context to Odd Thread Reporting Line'\n");
 | 
						|
 | 
						|
    /* clear odd cache line prior to pull operation */
 | 
						|
    memset(cl_pair, 0, sizeof(cl_pair));
 | 
						|
    get_nvp(qts, target_nvp, &nvp);
 | 
						|
    set_cl_pair(qts, &nvp, cl_pair);
 | 
						|
 | 
						|
    /* Read some values from TIMA that we expect to see in cacheline */
 | 
						|
    qw1w0 = get_tima32(qts, target_pir, TM_QW1_OS + TM_WORD0);
 | 
						|
    qw3w0 = get_tima32(qts, target_pir, TM_QW3_HV_PHYS + TM_WORD0);
 | 
						|
    qw1w2 = get_tima32(qts, target_pir, TM_QW1_OS + TM_WORD2);
 | 
						|
    qw2w2 = get_tima32(qts, target_pir, TM_QW2_HV_POOL + TM_WORD2);
 | 
						|
    qw3b8 = get_tima8(qts, target_pir, TM_QW3_HV_PHYS + TM_WORD2);
 | 
						|
 | 
						|
    /* Execute the pull operation */
 | 
						|
    set_tima8(qts, target_pir, TM_SPC_PULL_PHYS_CTX_OL, 0);
 | 
						|
 | 
						|
    /* Verify odd cache line values match TIMA after pull operation */
 | 
						|
    get_cl_pair(qts, &nvp, cl_pair);
 | 
						|
    memcpy(&cl_word, &cl_pair[XIVE_ODD_CL + TM_QW1_OS + TM_WORD0], 4);
 | 
						|
    g_assert_cmphex(qw1w0, ==, be32_to_cpu(cl_word));
 | 
						|
    memcpy(&cl_word, &cl_pair[XIVE_ODD_CL + TM_QW3_HV_PHYS + TM_WORD0], 4);
 | 
						|
    g_assert_cmphex(qw3w0, ==, be32_to_cpu(cl_word));
 | 
						|
    memcpy(&cl_word, &cl_pair[XIVE_ODD_CL + TM_QW1_OS + TM_WORD2], 4);
 | 
						|
    g_assert_cmphex(qw1w2, ==, be32_to_cpu(cl_word));
 | 
						|
    memcpy(&cl_word, &cl_pair[XIVE_ODD_CL + TM_QW2_HV_POOL + TM_WORD2], 4);
 | 
						|
    g_assert_cmphex(qw2w2, ==, be32_to_cpu(cl_word));
 | 
						|
    g_assert_cmphex(qw3b8, ==,
 | 
						|
                    cl_pair[XIVE_ODD_CL + TM_QW3_HV_PHYS + TM_WORD2]);
 | 
						|
 | 
						|
    /* Verify that all TIMA valid bits for target thread are cleared */
 | 
						|
    word2 = get_tima32(qts, target_pir, TM_QW1_OS + TM_WORD2);
 | 
						|
    g_assert_cmphex(xive_get_field32(TM_QW1W2_VO, word2), ==, 0);
 | 
						|
    word2 = get_tima32(qts, target_pir, TM_QW2_HV_POOL + TM_WORD2);
 | 
						|
    g_assert_cmphex(xive_get_field32(TM_QW2W2_VP, word2), ==, 0);
 | 
						|
    word2 = get_tima32(qts, target_pir, TM_QW3_HV_PHYS + TM_WORD2);
 | 
						|
    g_assert_cmphex(xive_get_field32(TM_QW3W2_VT, word2), ==, 0);
 | 
						|
}
 | 
						|
static void test_xive(void)
 | 
						|
{
 | 
						|
    QTestState *qts;
 | 
						|
 | 
						|
    qts = qtest_initf("-M powernv10 -smp %d,cores=1,threads=%d -nographic "
 | 
						|
                      "-nodefaults -serial mon:stdio -S "
 | 
						|
                      "-d guest_errors -trace '*xive*'",
 | 
						|
                      SMT, SMT);
 | 
						|
    init_xive(qts);
 | 
						|
 | 
						|
    test_hw_irq(qts);
 | 
						|
 | 
						|
    /* omit reset_state here and use settings from test_hw_irq */
 | 
						|
    test_pull_thread_ctx_to_odd_thread_cl(qts);
 | 
						|
 | 
						|
    reset_state(qts);
 | 
						|
    test_flush_sync_inject(qts);
 | 
						|
 | 
						|
    qtest_quit(qts);
 | 
						|
}
 | 
						|
 | 
						|
int main(int argc, char **argv)
 | 
						|
{
 | 
						|
    g_test_init(&argc, &argv, NULL);
 | 
						|
    qtest_add_func("xive2", test_xive);
 | 
						|
    return g_test_run();
 | 
						|
}
 |