From: Laszlo Ersek <lersek@redhat.com>
To: qemu-devel@nongnu.org, lersek@redhat.com
Cc: Marcel Apfelbaum <marcel@redhat.com>,
"Michael S. Tsirkin" <mst@redhat.com>
Subject: [Qemu-devel] [PATCH 3/4] hw/pci: allow the caller of pci_bar_address() to ignore command register
Date: Sat, 6 Jun 2015 01:46:28 +0200 [thread overview]
Message-ID: <1433547989-7238-4-git-send-email-lersek@redhat.com> (raw)
In-Reply-To: <1433547989-7238-1-git-send-email-lersek@redhat.com>
The pci_bar_address() function calculates the start address of a given PCI
BAR. Its main purpose is supporting MemoryRegion mapping / unmapping in
pci_update_mappings(), which is guest-observable.
For that reason it contains heavy logic to enforce the validity of the
BAR, and this logic includes checks for the PCI command register:
- If the PCI_COMMAND_IO bit is clear in the command register, then all IO
space BARs will be reported unmapped (all bits one),
- If the PCI_COMMAND_MEMORY bit is clear in the command register, then all
MMIO space BARs will be reported unmapped (all bits one).
For an upcoming use case, we'd like pci_bar_address() to calculate the BAR
address from the config space *regardless* of the command register. Add a
new parameter that allows the caller to cut out the command register
verification.
All current callers preserve their behaviors.
Cc: Marcel Apfelbaum <marcel@redhat.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
---
include/hw/pci/pci.h | 2 ++
hw/pci/pci.c | 18 +++++++++++-------
2 files changed, 13 insertions(+), 7 deletions(-)
diff --git a/include/hw/pci/pci.h b/include/hw/pci/pci.h
index d44bc84..16b467b 100644
--- a/include/hw/pci/pci.h
+++ b/include/hw/pci/pci.h
@@ -314,6 +314,8 @@ void pci_register_vga(PCIDevice *pci_dev, MemoryRegion *mem,
MemoryRegion *io_lo, MemoryRegion *io_hi);
void pci_unregister_vga(PCIDevice *pci_dev);
pcibus_t pci_get_bar_addr(PCIDevice *pci_dev, int region_num);
+pcibus_t pci_bar_address(PCIDevice *d, int reg, uint8_t type, pcibus_t size,
+ bool respect_command);
int pci_add_capability(PCIDevice *pdev, uint8_t cap_id,
uint8_t offset, uint8_t size);
diff --git a/hw/pci/pci.c b/hw/pci/pci.c
index 750f3da..61a70de 100644
--- a/hw/pci/pci.c
+++ b/hw/pci/pci.c
@@ -1054,15 +1054,19 @@ pcibus_t pci_get_bar_addr(PCIDevice *pci_dev, int region_num)
return pci_dev->io_regions[region_num].addr;
}
-static pcibus_t pci_bar_address(PCIDevice *d,
- int reg, uint8_t type, pcibus_t size)
+pcibus_t pci_bar_address(PCIDevice *d, int reg, uint8_t type, pcibus_t size,
+ bool respect_command)
{
pcibus_t new_addr, last_addr;
int bar = pci_bar(d, reg);
- uint16_t cmd = pci_get_word(d->config + PCI_COMMAND);
+ uint16_t cmd;
+
+ if (respect_command) {
+ cmd = pci_get_word(d->config + PCI_COMMAND);
+ }
if (type & PCI_BASE_ADDRESS_SPACE_IO) {
- if (!(cmd & PCI_COMMAND_IO)) {
+ if (respect_command && !(cmd & PCI_COMMAND_IO)) {
return PCI_BAR_UNMAPPED;
}
new_addr = pci_get_long(d->config + bar) & ~(size - 1);
@@ -1076,7 +1080,7 @@ static pcibus_t pci_bar_address(PCIDevice *d,
return new_addr;
}
- if (!(cmd & PCI_COMMAND_MEMORY)) {
+ if (respect_command && !(cmd & PCI_COMMAND_MEMORY)) {
return PCI_BAR_UNMAPPED;
}
if (type & PCI_BASE_ADDRESS_MEM_TYPE_64) {
@@ -1134,7 +1138,7 @@ static void pci_update_mappings(PCIDevice *d)
if (!r->size)
continue;
- new_addr = pci_bar_address(d, i, r->type, r->size);
+ new_addr = pci_bar_address(d, i, r->type, r->size, true);
/* This bar isn't changed */
if (new_addr == r->addr)
@@ -2427,7 +2431,7 @@ static void pci_dev_get_w64(PCIBus *b, PCIDevice *dev, void *opaque)
!(r->type & PCI_BASE_ADDRESS_MEM_TYPE_64)) {
continue;
}
- region_range.begin = pci_bar_address(dev, i, r->type, r->size);
+ region_range.begin = pci_bar_address(dev, i, r->type, r->size, true);
region_range.end = region_range.begin + r->size;
if (region_range.begin == PCI_BAR_UNMAPPED) {
--
1.8.3.1
next prev parent reply other threads:[~2015-06-05 23:46 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-06-05 23:45 [Qemu-devel] PXB fixes for QEMU, and extra root buses for OVMF Laszlo Ersek
2015-06-05 23:46 ` [Qemu-devel] [PATCH 0/4] PXB tweaks and fixes Laszlo Ersek
2015-06-05 23:46 ` [Qemu-devel] [PATCH 1/4] i386/acpi-build: more traditional _UID and _HID for PXB root buses Laszlo Ersek
2015-06-10 9:16 ` Marcel Apfelbaum
2015-06-05 23:46 ` [Qemu-devel] [PATCH 2/4] i386/acpi-build: fix PXB workarounds for unsupported BIOSes Laszlo Ersek
2015-06-10 9:17 ` Marcel Apfelbaum
2015-06-05 23:46 ` Laszlo Ersek [this message]
2015-06-05 23:46 ` [Qemu-devel] [PATCH 4/4] i386/acpi-build: build_crs(): fetch BAR from PCI config space directly Laszlo Ersek
2015-06-07 9:23 ` Michael S. Tsirkin
2015-06-08 7:56 ` Laszlo Ersek
2015-06-08 9:40 ` Michael S. Tsirkin
2015-06-09 20:34 ` Laszlo Ersek
2015-06-10 10:06 ` Marcel Apfelbaum
2015-06-10 11:07 ` Laszlo Ersek
2015-06-10 16:21 ` Michael S. Tsirkin
2015-06-10 16:19 ` Michael S. Tsirkin
2015-06-10 9:09 ` [Qemu-devel] PXB fixes for QEMU, and extra root buses for OVMF Marcel Apfelbaum
2015-06-10 11:04 ` Laszlo Ersek
2015-06-10 11:55 ` Marcel Apfelbaum
2015-06-10 12:05 ` Laszlo Ersek
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1433547989-7238-4-git-send-email-lersek@redhat.com \
--to=lersek@redhat.com \
--cc=marcel@redhat.com \
--cc=mst@redhat.com \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).