[Qemu-devel] [PATCH 0/5] spapr_pci: various cleanups and improvements

[Qemu-devel] [PATCH 0/5] spapr_pci: various cleanups and improvements

[Greg Kurz]

Before resuming the huge work on PHB hotplug, here are some patches
that maybe worth to apply.

Patches 1 to 3 are basic improvements.

Patch 4 and 5 may be a bit controversial. Everywhere in the spapr
code where we build an FDT portion, libfdt failures cause QEMU to
exit, even on hotplug paths. Only spapr_pci doesn't do that and
propagates the error instead. My understanding is that a failure
when building the FDT is likely to happen because of a bug in QEMU.

Hence the choice to convert spapr_pci to do like the others. We may
even consider changing _FDT() to abort() instead of exit().

Alternatively, if libfdt failures shouldn't be necessarily fatal,
especially on post-realize paths, then we should probably introduce
an _FDT_ERR() helper to propagate errors. And use it in may places
where we currently terminate QEMU: memory hotplug, CPU hotplug, CAS,
machine reset...

--
Greg

---

Greg Kurz (5):
      spapr_pci: drop useless check in spapr_phb_vfio_get_loc_code()
      spapr_pci: drop useless check in spapr_populate_pci_child_dt()
      spapr_pci: use g_strdup_printf()
      spapr_pci: use the common _FDT() helper
      spapr_pci: handle FDT creation errors with _FDT()


 hw/ppc/spapr_pci.c |   69 ++++++++++++++++------------------------------------
 1 file changed, 21 insertions(+), 48 deletions(-)

[Qemu-devel] [PATCH 1/5] spapr_pci: drop useless check in spapr_phb_vfio_get_loc_code()

[Greg Kurz]

g_strdup_printf() either returns a non-null pointer, either aborts
if it failed to allocate memory.

Signed-off-by: Greg Kurz <groug@kaod.org>
---
 hw/ppc/spapr_pci.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/hw/ppc/spapr_pci.c b/hw/ppc/spapr_pci.c
index d7880f257aa1..ef982f2ef370 100644
--- a/hw/ppc/spapr_pci.c
+++ b/hw/ppc/spapr_pci.c
@@ -766,7 +766,7 @@ static char *spapr_phb_vfio_get_loc_code(sPAPRPHBState *sphb,  PCIDevice *pdev)
     /* Construct the path of the file that will give us the DT location */
     path = g_strdup_printf("/sys/bus/pci/devices/%s/devspec", host);
     g_free(host);
-    if (!path || !g_file_get_contents(path, &buf, NULL, NULL)) {
+    if (!g_file_get_contents(path, &buf, NULL, NULL)) {
         goto err_out;
     }
     g_free(path);
@@ -774,7 +774,7 @@ static char *spapr_phb_vfio_get_loc_code(sPAPRPHBState *sphb,  PCIDevice *pdev)
     /* Construct and read from host device tree the loc-code */
     path = g_strdup_printf("/proc/device-tree%s/ibm,loc-code", buf);
     g_free(buf);
-    if (!path || !g_file_get_contents(path, &buf, NULL, NULL)) {
+    if (!g_file_get_contents(path, &buf, NULL, NULL)) {
         goto err_out;
     }
     return buf;

Re: [Qemu-devel] [PATCH 1/5] spapr_pci: drop useless check in spapr_phb_vfio_get_loc_code()

[Philippe Mathieu-Daudé]

On 09/09/2017 12:06 PM, Greg Kurz wrote:
> g_strdup_printf() either returns a non-null pointer, either aborts
> if it failed to allocate memory.
> 
> Signed-off-by: Greg Kurz <groug@kaod.org>

Acked-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

> ---
>   hw/ppc/spapr_pci.c |    4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/hw/ppc/spapr_pci.c b/hw/ppc/spapr_pci.c
> index d7880f257aa1..ef982f2ef370 100644
> --- a/hw/ppc/spapr_pci.c
> +++ b/hw/ppc/spapr_pci.c
> @@ -766,7 +766,7 @@ static char *spapr_phb_vfio_get_loc_code(sPAPRPHBState *sphb,  PCIDevice *pdev)
>       /* Construct the path of the file that will give us the DT location */
>       path = g_strdup_printf("/sys/bus/pci/devices/%s/devspec", host);
>       g_free(host);
> -    if (!path || !g_file_get_contents(path, &buf, NULL, NULL)) {
> +    if (!g_file_get_contents(path, &buf, NULL, NULL)) {
>           goto err_out;
>       }
>       g_free(path);
> @@ -774,7 +774,7 @@ static char *spapr_phb_vfio_get_loc_code(sPAPRPHBState *sphb,  PCIDevice *pdev)
>       /* Construct and read from host device tree the loc-code */
>       path = g_strdup_printf("/proc/device-tree%s/ibm,loc-code", buf);
>       g_free(buf);
> -    if (!path || !g_file_get_contents(path, &buf, NULL, NULL)) {
> +    if (!g_file_get_contents(path, &buf, NULL, NULL)) {
>           goto err_out;
>       }
>       return buf;
> 
> 

[Qemu-devel] [PATCH 2/5] spapr_pci: drop useless check in spapr_populate_pci_child_dt()

[Greg Kurz]

spapr_phb_get_loc_code() either returns a non-null pointer, either
aborts if g_strdup_printf() failed to allocate memory.

Signed-off-by: Greg Kurz <groug@kaod.org>
---
 hw/ppc/spapr_pci.c |    6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/hw/ppc/spapr_pci.c b/hw/ppc/spapr_pci.c
index ef982f2ef370..cd8efb181223 100644
--- a/hw/ppc/spapr_pci.c
+++ b/hw/ppc/spapr_pci.c
@@ -1282,12 +1282,8 @@ static int spapr_populate_pci_child_dt(PCIDevice *dev, void *fdt, int offset,
                             pci_find_device_name((ccode >> 16) & 0xff,
                                                  (ccode >> 8) & 0xff,
                                                  ccode & 0xff)));
-    buf = spapr_phb_get_loc_code(sphb, dev);
-    if (!buf) {
-        error_report("Failed setting the ibm,loc-code");
-        return -1;
-    }
 
+    buf = spapr_phb_get_loc_code(sphb, dev);
     err = fdt_setprop_string(fdt, offset, "ibm,loc-code", buf);
     g_free(buf);
     if (err < 0) {

Re: [Qemu-devel] [PATCH 2/5] spapr_pci: drop useless check in spapr_populate_pci_child_dt()

[Philippe Mathieu-Daudé]

On 09/09/2017 12:06 PM, Greg Kurz wrote:
> spapr_phb_get_loc_code() either returns a non-null pointer, either
> aborts if g_strdup_printf() failed to allocate memory.
> 
> Signed-off-by: Greg Kurz <groug@kaod.org>

Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

> ---
>   hw/ppc/spapr_pci.c |    6 +-----
>   1 file changed, 1 insertion(+), 5 deletions(-)
> 
> diff --git a/hw/ppc/spapr_pci.c b/hw/ppc/spapr_pci.c
> index ef982f2ef370..cd8efb181223 100644
> --- a/hw/ppc/spapr_pci.c
> +++ b/hw/ppc/spapr_pci.c
> @@ -1282,12 +1282,8 @@ static int spapr_populate_pci_child_dt(PCIDevice *dev, void *fdt, int offset,
>                               pci_find_device_name((ccode >> 16) & 0xff,
>                                                    (ccode >> 8) & 0xff,
>                                                    ccode & 0xff)));
> -    buf = spapr_phb_get_loc_code(sphb, dev);
> -    if (!buf) {
> -        error_report("Failed setting the ibm,loc-code");
> -        return -1;
> -    }
>   
> +    buf = spapr_phb_get_loc_code(sphb, dev);
>       err = fdt_setprop_string(fdt, offset, "ibm,loc-code", buf);
>       g_free(buf);
>       if (err < 0) {
> 
> 

[Qemu-devel] [PATCH 3/5] spapr_pci: use g_strdup_printf()

[Greg Kurz]

Building strings with g_strdup_printf() instead of snprintf() is
a QEMU common practice.

Signed-off-by: Greg Kurz <groug@kaod.org>
---
 hw/ppc/spapr_pci.c |   19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)

diff --git a/hw/ppc/spapr_pci.c b/hw/ppc/spapr_pci.c
index cd8efb181223..6da73fe6bc29 100644
--- a/hw/ppc/spapr_pci.c
+++ b/hw/ppc/spapr_pci.c
@@ -61,8 +61,6 @@
 #define RTAS_TYPE_MSI           1
 #define RTAS_TYPE_MSIX          2
 
-#define FDT_NAME_MAX          128
-
 #define _FDT(exp) \
     do { \
         int ret = (exp);                                           \
@@ -1194,7 +1192,7 @@ static const char *pci_find_device_name(uint8_t class, uint8_t subclass,
     return name;
 }
 
-static void pci_get_node_name(char *nodename, int len, PCIDevice *dev)
+static gchar *pci_get_node_name(PCIDevice *dev)
 {
     int slot = PCI_SLOT(dev->devfn);
     int func = PCI_FUNC(dev->devfn);
@@ -1205,9 +1203,9 @@ static void pci_get_node_name(char *nodename, int len, PCIDevice *dev)
                                 ccode & 0xff);
 
     if (func != 0) {
-        snprintf(nodename, len, "%s@%x,%x", name, slot, func);
+        return g_strdup_printf("%s@%x,%x", name, slot, func);
     } else {
-        snprintf(nodename, len, "%s@%x", name, slot);
+        return g_strdup_printf("%s@%x", name, slot);
     }
 }
 
@@ -1325,10 +1323,12 @@ static int spapr_create_pci_child_dt(sPAPRPHBState *phb, PCIDevice *dev,
                                      void *fdt, int node_offset)
 {
     int offset, ret;
-    char nodename[FDT_NAME_MAX];
+    gchar *nodename;
 
-    pci_get_node_name(nodename, FDT_NAME_MAX, dev);
+    nodename = pci_get_node_name(dev);
     offset = fdt_add_subnode(fdt, node_offset, nodename);
+    g_free(nodename);
+
     ret = spapr_populate_pci_child_dt(dev, fdt, offset, phb);
 
     g_assert(!ret);
@@ -2072,7 +2072,7 @@ int spapr_populate_pci_dt(sPAPRPHBState *phb,
                           void *fdt)
 {
     int bus_off, i, j, ret;
-    char nodename[FDT_NAME_MAX];
+    gchar *nodename;
     uint32_t bus_range[] = { cpu_to_be32(0), cpu_to_be32(0xff) };
     struct {
         uint32_t hi;
@@ -2121,8 +2121,9 @@ int spapr_populate_pci_dt(sPAPRPHBState *phb,
     sPAPRFDT s_fdt;
 
     /* Start populating the FDT */
-    snprintf(nodename, FDT_NAME_MAX, "pci@%" PRIx64, phb->buid);
+    nodename = g_strdup_printf("pci@%" PRIx64, phb->buid);
     bus_off = fdt_add_subnode(fdt, 0, nodename);
+    g_free(nodename);
     if (bus_off < 0) {
         return bus_off;
     }

[Qemu-devel] [PATCH 4/5] spapr_pci: use the common _FDT() helper

[Greg Kurz]

All other users in hw/ppc already consider an error when building
the FDT to be fatal, even on hotplug paths. There's no valid reason
for spapr_pci to behave differently. So let's used the common _FDT()
helper which terminates QEMU when libfdt fails.

Signed-off-by: Greg Kurz <groug@kaod.org>
---
 hw/ppc/spapr_pci.c |   10 +---------
 1 file changed, 1 insertion(+), 9 deletions(-)

diff --git a/hw/ppc/spapr_pci.c b/hw/ppc/spapr_pci.c
index 6da73fe6bc29..abb9f05e7b75 100644
--- a/hw/ppc/spapr_pci.c
+++ b/hw/ppc/spapr_pci.c
@@ -40,7 +40,7 @@
 #include "trace.h"
 #include "qemu/error-report.h"
 #include "qapi/qmp/qerror.h"
-
+#include "hw/ppc/fdt.h"
 #include "hw/pci/pci_bridge.h"
 #include "hw/pci/pci_bus.h"
 #include "hw/pci/pci_ids.h"
@@ -61,14 +61,6 @@
 #define RTAS_TYPE_MSI           1
 #define RTAS_TYPE_MSIX          2
 
-#define _FDT(exp) \
-    do { \
-        int ret = (exp);                                           \
-        if (ret < 0) {                                             \
-            return ret;                                            \
-        }                                                          \
-    } while (0)
-
 sPAPRPHBState *spapr_pci_find_phb(sPAPRMachineState *spapr, uint64_t buid)
 {
     sPAPRPHBState *sphb;

[Qemu-devel] [PATCH 5/5] spapr_pci: handle FDT creation errors with _FDT()

[Greg Kurz]

libfdt failures when creating the FDT should cause QEMU to terminate.

Let's use the _FDT() macro which does just that instead of propagating
the error to the caller. spapr_populate_pci_child_dt() no longer needs
to return a value in this case.

Note that, on the way, this get rids of the following nonsensical lines:

    g_assert(!ret);
    if (ret) {

Signed-off-by: Greg Kurz <groug@kaod.org>
---
 hw/ppc/spapr_pci.c |   30 +++++++-----------------------
 1 file changed, 7 insertions(+), 23 deletions(-)

diff --git a/hw/ppc/spapr_pci.c b/hw/ppc/spapr_pci.c
index abb9f05e7b75..75cd9392233e 100644
--- a/hw/ppc/spapr_pci.c
+++ b/hw/ppc/spapr_pci.c
@@ -1204,12 +1204,12 @@ static gchar *pci_get_node_name(PCIDevice *dev)
 static uint32_t spapr_phb_get_pci_drc_index(sPAPRPHBState *phb,
                                             PCIDevice *pdev);
 
-static int spapr_populate_pci_child_dt(PCIDevice *dev, void *fdt, int offset,
+static void spapr_populate_pci_child_dt(PCIDevice *dev, void *fdt, int offset,
                                        sPAPRPHBState *sphb)
 {
     ResourceProps rp;
     bool is_bridge = false;
-    int pci_status, err;
+    int pci_status;
     char *buf = NULL;
     uint32_t drc_index = spapr_phb_get_pci_drc_index(sphb, dev);
     uint32_t ccode = pci_default_read_config(dev, PCI_CLASS_PROG, 3);
@@ -1274,11 +1274,8 @@ static int spapr_populate_pci_child_dt(PCIDevice *dev, void *fdt, int offset,
                                                  ccode & 0xff)));
 
     buf = spapr_phb_get_loc_code(sphb, dev);
-    err = fdt_setprop_string(fdt, offset, "ibm,loc-code", buf);
+    _FDT(fdt_setprop_string(fdt, offset, "ibm,loc-code", buf));
     g_free(buf);
-    if (err < 0) {
-        return err;
-    }
 
     if (drc_index) {
         _FDT(fdt_setprop_cell(fdt, offset, "ibm,my-drc-index", drc_index));
@@ -1306,27 +1303,21 @@ static int spapr_populate_pci_child_dt(PCIDevice *dev, void *fdt, int offset,
     if (sphb->pcie_ecs && pci_is_express(dev)) {
         _FDT(fdt_setprop_cell(fdt, offset, "ibm,pci-config-space-type", 0x1));
     }
-
-    return 0;
 }
 
 /* create OF node for pci device and required OF DT properties */
 static int spapr_create_pci_child_dt(sPAPRPHBState *phb, PCIDevice *dev,
                                      void *fdt, int node_offset)
 {
-    int offset, ret;
+    int offset;
     gchar *nodename;
 
     nodename = pci_get_node_name(dev);
-    offset = fdt_add_subnode(fdt, node_offset, nodename);
+    _FDT(offset = fdt_add_subnode(fdt, node_offset, nodename));
     g_free(nodename);
 
-    ret = spapr_populate_pci_child_dt(dev, fdt, offset, phb);
+    spapr_populate_pci_child_dt(dev, fdt, offset, phb);
 
-    g_assert(!ret);
-    if (ret) {
-        return 0;
-    }
     return offset;
 }
 
@@ -1416,10 +1407,6 @@ static void spapr_pci_plug(HotplugHandler *plug_handler,
 
     fdt = create_device_tree(&fdt_size);
     fdt_start_offset = spapr_create_pci_child_dt(phb, pdev, fdt, 0);
-    if (!fdt_start_offset) {
-        error_setg(&local_err, "Failed to create pci child device tree node");
-        goto out;
-    }
 
     spapr_drc_attach(drc, DEVICE(pdev), fdt, fdt_start_offset, &local_err);
     if (local_err) {
@@ -2114,11 +2101,8 @@ int spapr_populate_pci_dt(sPAPRPHBState *phb,
 
     /* Start populating the FDT */
     nodename = g_strdup_printf("pci@%" PRIx64, phb->buid);
-    bus_off = fdt_add_subnode(fdt, 0, nodename);
+    _FDT(bus_off = fdt_add_subnode(fdt, 0, nodename));
     g_free(nodename);
-    if (bus_off < 0) {
-        return bus_off;
-    }
 
     /* Write PHB properties */
     _FDT(fdt_setprop_string(fdt, bus_off, "device_type", "pci"));

Re: [Qemu-devel] [PATCH 0/5] spapr_pci: various cleanups and improvements

[David Gibson]

[-- Attachment #1: Type: text/plain, Size: 1265 bytes --]

On Sat, Sep 09, 2017 at 05:05:46PM +0200, Greg Kurz wrote:
> Before resuming the huge work on PHB hotplug, here are some patches
> that maybe worth to apply.
> 
> Patches 1 to 3 are basic improvements.

I've applied these to ppc-for-2.11.

> Patch 4 and 5 may be a bit controversial. Everywhere in the spapr
> code where we build an FDT portion, libfdt failures cause QEMU to
> exit, even on hotplug paths. Only spapr_pci doesn't do that and
> propagates the error instead. My understanding is that a failure
> when building the FDT is likely to happen because of a bug in QEMU.

Still looking at these.

> 
> Hence the choice to convert spapr_pci to do like the others. We may
> even consider changing _FDT() to abort() instead of exit().
> 
> Alternatively, if libfdt failures shouldn't be necessarily fatal,
> especially on post-realize paths, then we should probably introduce
> an _FDT_ERR() helper to propagate errors. And use it in may places
> where we currently terminate QEMU: memory hotplug, CPU hotplug, CAS,
> machine reset...
> 

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [Qemu-devel] [PATCH 0/5] spapr_pci: various cleanups and improvements

[David Gibson]

[-- Attachment #1: Type: text/plain, Size: 1418 bytes --]

On Sat, Sep 09, 2017 at 05:05:46PM +0200, Greg Kurz wrote:
> Before resuming the huge work on PHB hotplug, here are some patches
> that maybe worth to apply.
> 
> Patches 1 to 3 are basic improvements.
> 
> Patch 4 and 5 may be a bit controversial. Everywhere in the spapr
> code where we build an FDT portion, libfdt failures cause QEMU to
> exit, even on hotplug paths. Only spapr_pci doesn't do that and
> propagates the error instead. My understanding is that a failure
> when building the FDT is likely to happen because of a bug in QEMU.
> 
> Hence the choice to convert spapr_pci to do like the others. We may
> even consider changing _FDT() to abort() instead of exit().
> 
> Alternatively, if libfdt failures shouldn't be necessarily fatal,
> especially on post-realize paths, then we should probably introduce
> an _FDT_ERR() helper to propagate errors. And use it in may places
> where we currently terminate QEMU: memory hotplug, CPU hotplug, CAS,
> machine reset...

I've applied 4 & 5 as well now.  If I get to it, I hope to make them
obsolete by replacing most of the fdt wrangling with a qemu internal
live tree representation, but in the meantime it's still a reasonable
cleanup.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]