mirror of
https://github.com/AuxXxilium/linux_dsm_epyc7002.git
synced 2024-12-27 18:25:04 +07:00
d6776bba44
Keep IRQ mappings on context teardown. This won't leak IRQs as if we allocate the mapping again, the generic code will give the same mapping used last time. Doing this works around a race in the generic code. Masking the interrupt introduces a race which can crash the kernel or result in IRQ that is never EOIed. The lost of EOI results in all subsequent mappings to the same HW IRQ never receiving an interrupt. We've seen this race with cxl test cases which are doing heavy context startup and teardown at the same time as heavy interrupt load. A fix to the generic code is being investigated also. Signed-off-by: Michael Neuling <mikey@neuling.org> Cc: stable@vger.kernel.org # 3.8 Tested-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com> Acked-by: Ian Munsie <imunsie@au1.ibm.com> Tested-by: Vaibhav Jain <vaibhav@linux.vnet.ibm.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
377 lines
10 KiB
C
377 lines
10 KiB
C
/*
|
|
* Copyright 2014 IBM Corp.
|
|
*
|
|
* This program is free software; you can redistribute it and/or
|
|
* modify it under the terms of the GNU General Public License
|
|
* as published by the Free Software Foundation; either version
|
|
* 2 of the License, or (at your option) any later version.
|
|
*/
|
|
|
|
#include <linux/interrupt.h>
|
|
#include <linux/workqueue.h>
|
|
#include <linux/sched.h>
|
|
#include <linux/wait.h>
|
|
#include <linux/slab.h>
|
|
#include <linux/pid.h>
|
|
#include <asm/cputable.h>
|
|
#include <misc/cxl-base.h>
|
|
|
|
#include "cxl.h"
|
|
#include "trace.h"
|
|
|
|
static int afu_irq_range_start(void)
|
|
{
|
|
if (cpu_has_feature(CPU_FTR_HVMODE))
|
|
return 1;
|
|
return 0;
|
|
}
|
|
|
|
static irqreturn_t schedule_cxl_fault(struct cxl_context *ctx, u64 dsisr, u64 dar)
|
|
{
|
|
ctx->dsisr = dsisr;
|
|
ctx->dar = dar;
|
|
schedule_work(&ctx->fault_work);
|
|
return IRQ_HANDLED;
|
|
}
|
|
|
|
irqreturn_t cxl_irq(int irq, struct cxl_context *ctx, struct cxl_irq_info *irq_info)
|
|
{
|
|
u64 dsisr, dar;
|
|
|
|
dsisr = irq_info->dsisr;
|
|
dar = irq_info->dar;
|
|
|
|
trace_cxl_psl_irq(ctx, irq, dsisr, dar);
|
|
|
|
pr_devel("CXL interrupt %i for afu pe: %i DSISR: %#llx DAR: %#llx\n", irq, ctx->pe, dsisr, dar);
|
|
|
|
if (dsisr & CXL_PSL_DSISR_An_DS) {
|
|
/*
|
|
* We don't inherently need to sleep to handle this, but we do
|
|
* need to get a ref to the task's mm, which we can't do from
|
|
* irq context without the potential for a deadlock since it
|
|
* takes the task_lock. An alternate option would be to keep a
|
|
* reference to the task's mm the entire time it has cxl open,
|
|
* but to do that we need to solve the issue where we hold a
|
|
* ref to the mm, but the mm can hold a ref to the fd after an
|
|
* mmap preventing anything from being cleaned up.
|
|
*/
|
|
pr_devel("Scheduling segment miss handling for later pe: %i\n", ctx->pe);
|
|
return schedule_cxl_fault(ctx, dsisr, dar);
|
|
}
|
|
|
|
if (dsisr & CXL_PSL_DSISR_An_M)
|
|
pr_devel("CXL interrupt: PTE not found\n");
|
|
if (dsisr & CXL_PSL_DSISR_An_P)
|
|
pr_devel("CXL interrupt: Storage protection violation\n");
|
|
if (dsisr & CXL_PSL_DSISR_An_A)
|
|
pr_devel("CXL interrupt: AFU lock access to write through or cache inhibited storage\n");
|
|
if (dsisr & CXL_PSL_DSISR_An_S)
|
|
pr_devel("CXL interrupt: Access was afu_wr or afu_zero\n");
|
|
if (dsisr & CXL_PSL_DSISR_An_K)
|
|
pr_devel("CXL interrupt: Access not permitted by virtual page class key protection\n");
|
|
|
|
if (dsisr & CXL_PSL_DSISR_An_DM) {
|
|
/*
|
|
* In some cases we might be able to handle the fault
|
|
* immediately if hash_page would succeed, but we still need
|
|
* the task's mm, which as above we can't get without a lock
|
|
*/
|
|
pr_devel("Scheduling page fault handling for later pe: %i\n", ctx->pe);
|
|
return schedule_cxl_fault(ctx, dsisr, dar);
|
|
}
|
|
if (dsisr & CXL_PSL_DSISR_An_ST)
|
|
WARN(1, "CXL interrupt: Segment Table PTE not found\n");
|
|
if (dsisr & CXL_PSL_DSISR_An_UR)
|
|
pr_devel("CXL interrupt: AURP PTE not found\n");
|
|
if (dsisr & CXL_PSL_DSISR_An_PE)
|
|
return cxl_ops->handle_psl_slice_error(ctx, dsisr,
|
|
irq_info->errstat);
|
|
if (dsisr & CXL_PSL_DSISR_An_AE) {
|
|
pr_devel("CXL interrupt: AFU Error 0x%016llx\n", irq_info->afu_err);
|
|
|
|
if (ctx->pending_afu_err) {
|
|
/*
|
|
* This shouldn't happen - the PSL treats these errors
|
|
* as fatal and will have reset the AFU, so there's not
|
|
* much point buffering multiple AFU errors.
|
|
* OTOH if we DO ever see a storm of these come in it's
|
|
* probably best that we log them somewhere:
|
|
*/
|
|
dev_err_ratelimited(&ctx->afu->dev, "CXL AFU Error "
|
|
"undelivered to pe %i: 0x%016llx\n",
|
|
ctx->pe, irq_info->afu_err);
|
|
} else {
|
|
spin_lock(&ctx->lock);
|
|
ctx->afu_err = irq_info->afu_err;
|
|
ctx->pending_afu_err = 1;
|
|
spin_unlock(&ctx->lock);
|
|
|
|
wake_up_all(&ctx->wq);
|
|
}
|
|
|
|
cxl_ops->ack_irq(ctx, CXL_PSL_TFC_An_A, 0);
|
|
return IRQ_HANDLED;
|
|
}
|
|
if (dsisr & CXL_PSL_DSISR_An_OC)
|
|
pr_devel("CXL interrupt: OS Context Warning\n");
|
|
|
|
WARN(1, "Unhandled CXL PSL IRQ\n");
|
|
return IRQ_HANDLED;
|
|
}
|
|
|
|
static irqreturn_t cxl_irq_afu(int irq, void *data)
|
|
{
|
|
struct cxl_context *ctx = data;
|
|
irq_hw_number_t hwirq = irqd_to_hwirq(irq_get_irq_data(irq));
|
|
int irq_off, afu_irq = 0;
|
|
__u16 range;
|
|
int r;
|
|
|
|
/*
|
|
* Look for the interrupt number.
|
|
* On bare-metal, we know range 0 only contains the PSL
|
|
* interrupt so we could start counting at range 1 and initialize
|
|
* afu_irq at 1.
|
|
* In a guest, range 0 also contains AFU interrupts, so it must
|
|
* be counted for. Therefore we initialize afu_irq at 0 to take into
|
|
* account the PSL interrupt.
|
|
*
|
|
* For code-readability, it just seems easier to go over all
|
|
* the ranges on bare-metal and guest. The end result is the same.
|
|
*/
|
|
for (r = 0; r < CXL_IRQ_RANGES; r++) {
|
|
irq_off = hwirq - ctx->irqs.offset[r];
|
|
range = ctx->irqs.range[r];
|
|
if (irq_off >= 0 && irq_off < range) {
|
|
afu_irq += irq_off;
|
|
break;
|
|
}
|
|
afu_irq += range;
|
|
}
|
|
if (unlikely(r >= CXL_IRQ_RANGES)) {
|
|
WARN(1, "Received AFU IRQ out of range for pe %i (virq %i hwirq %lx)\n",
|
|
ctx->pe, irq, hwirq);
|
|
return IRQ_HANDLED;
|
|
}
|
|
|
|
trace_cxl_afu_irq(ctx, afu_irq, irq, hwirq);
|
|
pr_devel("Received AFU interrupt %i for pe: %i (virq %i hwirq %lx)\n",
|
|
afu_irq, ctx->pe, irq, hwirq);
|
|
|
|
if (unlikely(!ctx->irq_bitmap)) {
|
|
WARN(1, "Received AFU IRQ for context with no IRQ bitmap\n");
|
|
return IRQ_HANDLED;
|
|
}
|
|
spin_lock(&ctx->lock);
|
|
set_bit(afu_irq - 1, ctx->irq_bitmap);
|
|
ctx->pending_irq = true;
|
|
spin_unlock(&ctx->lock);
|
|
|
|
wake_up_all(&ctx->wq);
|
|
|
|
return IRQ_HANDLED;
|
|
}
|
|
|
|
unsigned int cxl_map_irq(struct cxl *adapter, irq_hw_number_t hwirq,
|
|
irq_handler_t handler, void *cookie, const char *name)
|
|
{
|
|
unsigned int virq;
|
|
int result;
|
|
|
|
/* IRQ Domain? */
|
|
virq = irq_create_mapping(NULL, hwirq);
|
|
if (!virq) {
|
|
dev_warn(&adapter->dev, "cxl_map_irq: irq_create_mapping failed\n");
|
|
return 0;
|
|
}
|
|
|
|
if (cxl_ops->setup_irq)
|
|
cxl_ops->setup_irq(adapter, hwirq, virq);
|
|
|
|
pr_devel("hwirq %#lx mapped to virq %u\n", hwirq, virq);
|
|
|
|
result = request_irq(virq, handler, 0, name, cookie);
|
|
if (result) {
|
|
dev_warn(&adapter->dev, "cxl_map_irq: request_irq failed: %i\n", result);
|
|
return 0;
|
|
}
|
|
|
|
return virq;
|
|
}
|
|
|
|
void cxl_unmap_irq(unsigned int virq, void *cookie)
|
|
{
|
|
free_irq(virq, cookie);
|
|
}
|
|
|
|
int cxl_register_one_irq(struct cxl *adapter,
|
|
irq_handler_t handler,
|
|
void *cookie,
|
|
irq_hw_number_t *dest_hwirq,
|
|
unsigned int *dest_virq,
|
|
const char *name)
|
|
{
|
|
int hwirq, virq;
|
|
|
|
if ((hwirq = cxl_ops->alloc_one_irq(adapter)) < 0)
|
|
return hwirq;
|
|
|
|
if (!(virq = cxl_map_irq(adapter, hwirq, handler, cookie, name)))
|
|
goto err;
|
|
|
|
*dest_hwirq = hwirq;
|
|
*dest_virq = virq;
|
|
|
|
return 0;
|
|
|
|
err:
|
|
cxl_ops->release_one_irq(adapter, hwirq);
|
|
return -ENOMEM;
|
|
}
|
|
|
|
void afu_irq_name_free(struct cxl_context *ctx)
|
|
{
|
|
struct cxl_irq_name *irq_name, *tmp;
|
|
|
|
list_for_each_entry_safe(irq_name, tmp, &ctx->irq_names, list) {
|
|
kfree(irq_name->name);
|
|
list_del(&irq_name->list);
|
|
kfree(irq_name);
|
|
}
|
|
}
|
|
|
|
int afu_allocate_irqs(struct cxl_context *ctx, u32 count)
|
|
{
|
|
int rc, r, i, j = 1;
|
|
struct cxl_irq_name *irq_name;
|
|
int alloc_count;
|
|
|
|
/*
|
|
* In native mode, range 0 is reserved for the multiplexed
|
|
* PSL interrupt. It has been allocated when the AFU was initialized.
|
|
*
|
|
* In a guest, the PSL interrupt is not mutliplexed, but per-context,
|
|
* and is the first interrupt from range 0. It still needs to be
|
|
* allocated, so bump the count by one.
|
|
*/
|
|
if (cpu_has_feature(CPU_FTR_HVMODE))
|
|
alloc_count = count;
|
|
else
|
|
alloc_count = count + 1;
|
|
|
|
/* Initialize the list head to hold irq names */
|
|
INIT_LIST_HEAD(&ctx->irq_names);
|
|
|
|
if ((rc = cxl_ops->alloc_irq_ranges(&ctx->irqs, ctx->afu->adapter,
|
|
alloc_count)))
|
|
return rc;
|
|
|
|
if (cpu_has_feature(CPU_FTR_HVMODE)) {
|
|
/* Multiplexed PSL Interrupt */
|
|
ctx->irqs.offset[0] = ctx->afu->native->psl_hwirq;
|
|
ctx->irqs.range[0] = 1;
|
|
}
|
|
|
|
ctx->irq_count = count;
|
|
ctx->irq_bitmap = kcalloc(BITS_TO_LONGS(count),
|
|
sizeof(*ctx->irq_bitmap), GFP_KERNEL);
|
|
if (!ctx->irq_bitmap)
|
|
goto out;
|
|
|
|
/*
|
|
* Allocate names first. If any fail, bail out before allocating
|
|
* actual hardware IRQs.
|
|
*/
|
|
for (r = afu_irq_range_start(); r < CXL_IRQ_RANGES; r++) {
|
|
for (i = 0; i < ctx->irqs.range[r]; i++) {
|
|
irq_name = kmalloc(sizeof(struct cxl_irq_name),
|
|
GFP_KERNEL);
|
|
if (!irq_name)
|
|
goto out;
|
|
irq_name->name = kasprintf(GFP_KERNEL, "cxl-%s-pe%i-%i",
|
|
dev_name(&ctx->afu->dev),
|
|
ctx->pe, j);
|
|
if (!irq_name->name) {
|
|
kfree(irq_name);
|
|
goto out;
|
|
}
|
|
/* Add to tail so next look get the correct order */
|
|
list_add_tail(&irq_name->list, &ctx->irq_names);
|
|
j++;
|
|
}
|
|
}
|
|
return 0;
|
|
|
|
out:
|
|
cxl_ops->release_irq_ranges(&ctx->irqs, ctx->afu->adapter);
|
|
afu_irq_name_free(ctx);
|
|
return -ENOMEM;
|
|
}
|
|
|
|
static void afu_register_hwirqs(struct cxl_context *ctx)
|
|
{
|
|
irq_hw_number_t hwirq;
|
|
struct cxl_irq_name *irq_name;
|
|
int r, i;
|
|
irqreturn_t (*handler)(int irq, void *data);
|
|
|
|
/* We've allocated all memory now, so let's do the irq allocations */
|
|
irq_name = list_first_entry(&ctx->irq_names, struct cxl_irq_name, list);
|
|
for (r = afu_irq_range_start(); r < CXL_IRQ_RANGES; r++) {
|
|
hwirq = ctx->irqs.offset[r];
|
|
for (i = 0; i < ctx->irqs.range[r]; hwirq++, i++) {
|
|
if (r == 0 && i == 0)
|
|
/*
|
|
* The very first interrupt of range 0 is
|
|
* always the PSL interrupt, but we only
|
|
* need to connect a handler for guests,
|
|
* because there's one PSL interrupt per
|
|
* context.
|
|
* On bare-metal, the PSL interrupt is
|
|
* multiplexed and was setup when the AFU
|
|
* was configured.
|
|
*/
|
|
handler = cxl_ops->psl_interrupt;
|
|
else
|
|
handler = cxl_irq_afu;
|
|
cxl_map_irq(ctx->afu->adapter, hwirq, handler, ctx,
|
|
irq_name->name);
|
|
irq_name = list_next_entry(irq_name, list);
|
|
}
|
|
}
|
|
}
|
|
|
|
int afu_register_irqs(struct cxl_context *ctx, u32 count)
|
|
{
|
|
int rc;
|
|
|
|
rc = afu_allocate_irqs(ctx, count);
|
|
if (rc)
|
|
return rc;
|
|
|
|
afu_register_hwirqs(ctx);
|
|
return 0;
|
|
}
|
|
|
|
void afu_release_irqs(struct cxl_context *ctx, void *cookie)
|
|
{
|
|
irq_hw_number_t hwirq;
|
|
unsigned int virq;
|
|
int r, i;
|
|
|
|
for (r = afu_irq_range_start(); r < CXL_IRQ_RANGES; r++) {
|
|
hwirq = ctx->irqs.offset[r];
|
|
for (i = 0; i < ctx->irqs.range[r]; hwirq++, i++) {
|
|
virq = irq_find_mapping(NULL, hwirq);
|
|
if (virq)
|
|
cxl_unmap_irq(virq, cookie);
|
|
}
|
|
}
|
|
|
|
afu_irq_name_free(ctx);
|
|
cxl_ops->release_irq_ranges(&ctx->irqs, ctx->afu->adapter);
|
|
|
|
ctx->irq_count = 0;
|
|
}
|