Skip to contentSkip to conversation

xl: convert vcpuid to signed in main_vcpupin()

PATCH Superseded Dario Faggioli 3 participants 8 messages Aug 20, 2014 – Sep 3, 2014

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 xl: convert vcpuid to signed in main_vcpupin()

No functional changes, it just looks more correct, considering that at some point in the function we assign -1 to it (and at some other later point we check for it to be -1), to signify 'all vcpus'.

While at it, fix a coding style nit and improve error reporting.

---
 tools/libxl/xl_cmdimpl.c |   14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
tools/libxl/xl_cmdimpl.c +7 -7
index f1c136a..a29a579 100644
@@ -4601,8 +4601,9 @@ int main_vcpupin(int argc, char **argv)
     libxl_vcpuinfo *vcpuinfo;
     libxl_bitmap cpumap_hard, cpumap_soft;;
     libxl_bitmap *soft = &cpumap_soft, *hard = &cpumap_hard;
-    uint32_t vcpuid, domid;
     const char *vcpu, *hard_str, *soft_str;
+    uint32_t domid;
+    long vcpuid;
     char *endptr;
     int opt, nb_cpu, nb_vcpu, rc = -1;
 
@@ -4619,10 +4620,10 @@ int main_vcpupin(int argc, char **argv)
     soft_str = (argc > optind+3) ? argv[optind+3] : NULL;
 
     /* Figure out with which vCPU we are dealing with */
-    vcpuid = strtoul(vcpu, &endptr, 10);
-    if (vcpu == endptr) {
+    vcpuid = strtol(vcpu, &endptr, 10);
+    if (vcpu == endptr || vcpuid < 0) {
         if (strcmp(vcpu, "all")) {
-            fprintf(stderr, "Error: Invalid argument.\n");
+            fprintf(stderr, "Error: Invalid argument %s as VCPU.\n", vcpu);
             goto out;
         }
         vcpuid = -1;
@@ -4688,12 +4689,11 @@ int main_vcpupin(int argc, char **argv)
 
     if (vcpuid != -1) {
         if (libxl_set_vcpuaffinity(ctx, domid, vcpuid, hard, soft)) {
-            fprintf(stderr, "Could not set affinity for vcpu `%u'.\n",
+            fprintf(stderr, "Could not set affinity for vcpu `%ld'.\n",
                     vcpuid);
             goto out;
         }
-    }
-    else {
+    } else {
         if (!(vcpuinfo = libxl_list_vcpu(ctx, domid, &nb_vcpu, &nb_cpu))) {
             fprintf(stderr, "libxl_list_vcpu failed.\n");
             goto out;
#1 lore raw
Quoted 3 lines · Dario Faggioli

No functional changes, it just looks more correct, considering that at some point in the function we assign -1 to it (and at some other later point we check for it to be -1),

I think that's more than just "looks more correct", it's just wrong to compare an unsigned to -1, or to assign it.

And there may well be a functional change since gcc is no longer allowed to assume that the number is +ve and therefore cannot discard some of these checks any more. We are probably just lucky that gcc isn't doing so already (and unlucky that it isn't generating a warning...)

Quoted 15 lines

to
signify 'all vcpus'.

While at it, fix a coding style nit and improve error reporting.

---
 tools/libxl/xl_cmdimpl.c |   14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
tools/libxl/xl_cmdimpl.c +0 -0
index f1c136a..a29a579 100644

tools/libxl/xl_cmdimpl.c int main_vcpupin(int argc, char **argv) · line 4606

Quoted 7 lines
tools/libxl/xl_cmdimpl.c +2 -1
@@ -4601,8 +4601,9 @@ int main_vcpupin(int argc, char **argv)
     libxl_vcpuinfo *vcpuinfo;
     libxl_bitmap cpumap_hard, cpumap_soft;;
     libxl_bitmap *soft = &cpumap_soft, *hard = &cpumap_hard;
-    uint32_t vcpuid, domid;
     const char *vcpu, *hard_str, *soft_str;
+    uint32_t domid;
+    long vcpuid;

I think an int would be sufficiently large for this.

Ian.

#2 lore raw
Quoted 5 lines · Ian Campbell
Quoted 3 lines · Dario Faggioli

No functional changes, it just looks more correct, considering that at some point in the function we assign -1 to it (and at some other later point we check for it to be -1),

Quoted 8 lines

I think that's more than just "looks more correct", it's just wrong to compare an unsigned to -1, or to assign it.

And there may well be a functional change since gcc is no longer allowed to assume that the number is +ve and therefore cannot discard some of these checks any more. We are probably just lucky that gcc isn't doing so already (and unlucky that it isn't generating a warning...)

Right, and thanks for all the details. :-)

Should I change the changelog, if I repost?

Quoted 5 lines
Quoted 5 lines
---
 tools/libxl/xl_cmdimpl.c |   14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
Quoted 15 lines

tools/libxl/xl_cmdimpl.c int main_vcpupin(int argc, char **argv) · line 4606

Quoted 12 lines
tools/libxl/xl_cmdimpl.c +2 -1
index f1c136a..a29a579 100644
@@ -4601,8 +4601,9 @@ int main_vcpupin(int argc, char **argv)
     libxl_vcpuinfo *vcpuinfo;
     libxl_bitmap cpumap_hard, cpumap_soft;;
     libxl_bitmap *soft = &cpumap_soft, *hard = &cpumap_hard;
-    uint32_t vcpuid, domid;
     const char *vcpu, *hard_str, *soft_str;
+    uint32_t domid;
+    long vcpuid;

I think an int would be sufficiently large for this.

The only reason why I used a long is that the function used to do the actual conversion is strtol(), which returns a long. Strictly speaking, if using an int, I think I should then be checking for the returned value to be within [INT_MIN, INT_MAX] and error out if it's not.

That being said, we can probably use int and leave the range checking alone, as, if someone tries to pin vCPU INT_MAX+1, he/she deserves all the bad that will happen, but since I was polishing the code... :-P

Just let me know if you want it to be and int, and if you think the range checking should be there, and I'll copy.

Thanks and regards,
Dario

Signature · 4 lines
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://about.me/dario.faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel

#3 lore raw
Quoted 9 lines · Dario Faggioli
Quoted 8 lines · Ian Campbell
Quoted 3 lines · Dario Faggioli

No functional changes, it just looks more correct, considering that at some point in the function we assign -1 to it (and at some other later point we check for it to be -1),

I think that's more than just "looks more correct", it's just wrong to compare an unsigned to -1, or to assign it.

Quoted 8 lines
Quoted 5 lines

And there may well be a functional change since gcc is no longer allowed to assume that the number is +ve and therefore cannot discard some of these checks any more. We are probably just lucky that gcc isn't doing so already (and unlucky that it isn't generating a warning...)

Right, and thanks for all the details. :-)

Should I change the changelog, if I repost?

Yes, please.

Quoted 21 lines
Quoted 6 lines
Quoted 6 lines
---
 tools/libxl/xl_cmdimpl.c |   14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
Quoted 15 lines

tools/libxl/xl_cmdimpl.c int main_vcpupin(int argc, char **argv) · line 4606

Quoted 12 lines
tools/libxl/xl_cmdimpl.c +2 -1
index f1c136a..a29a579 100644
@@ -4601,8 +4601,9 @@ int main_vcpupin(int argc, char **argv)
     libxl_vcpuinfo *vcpuinfo;
     libxl_bitmap cpumap_hard, cpumap_soft;;
     libxl_bitmap *soft = &cpumap_soft, *hard = &cpumap_hard;
-    uint32_t vcpuid, domid;
     const char *vcpu, *hard_str, *soft_str;
+    uint32_t domid;
+    long vcpuid;

I think an int would be sufficiently large for this.

Quoted 2 lines

The only reason why I used a long is that the function used to do the actual conversion is strtol(), which returns a long.

Can't it use atoi?

Ian.

#4 lore raw
Quoted 22 lines · Ian Campbell
Quoted 21 lines · Dario Faggioli
Quoted 21 lines
Quoted 11 lines
---
 tools/libxl/xl_cmdimpl.c |   14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
tools/libxl/xl_cmdimpl.c +0 -0
index f1c136a..a29a579 100644

tools/libxl/xl_cmdimpl.c int main_vcpupin(int argc, char **argv) · line 4606

Quoted 7 lines
tools/libxl/xl_cmdimpl.c +2 -1
@@ -4601,8 +4601,9 @@ int main_vcpupin(int argc, char **argv)
     libxl_vcpuinfo *vcpuinfo;
     libxl_bitmap cpumap_hard, cpumap_soft;;
     libxl_bitmap *soft = &cpumap_soft, *hard = &cpumap_hard;
-    uint32_t vcpuid, domid;
     const char *vcpu, *hard_str, *soft_str;
+    uint32_t domid;
+    long vcpuid;

I think an int would be sufficiently large for this.

Quoted 5 lines
Quoted 2 lines

The only reason why I used a long is that the function used to do the actual conversion is strtol(), which returns a long.

Can't it use atoi?

From atoi(3):

"The behavior is the same as strtol(nptr, NULL, 10); except that atoi() does not detect errors."

While the current code uses some of the strtol() error handling capabilities:

    /* Figure out with which vCPU we are dealing with */                                                              
    vcpuid = strtoul(vcpu, &endptr, 10);                                                                              
    if (vcpu == endptr) {                                                                                             
        if (strcmp(vcpu, "all")) {                                                                                    
            fprintf(stderr, "Error: Invalid argument.\n");                                                            
            goto out;                                                                                                 
        }                                                                                                             
        vcpuid = -1;                                                                                                  
    }

I can try to reorganize the code a bit but, without being sure that the passed string, if not the string "all", is an actual number (which, AFAIUI, is what strtol() gives me that atoi() doesn't), I don't think we can be as precise in error detection/reporting as we are now (I just double checked, and i=atoi("foo") means i becomes 0).

I can probably combine atoi() and (something like) isdigit(vcpu[0]) to get to something similar, but at that point, I think I like strtol() better.

Let me know.

Regards,
Dario

Signature · 4 lines
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://about.me/dario.faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel

#5 lore raw
Quoted 52 lines · Dario Faggioli
Quoted 22 lines · Ian Campbell
Quoted 21 lines · Dario Faggioli
Quoted 21 lines
Quoted 11 lines
---
 tools/libxl/xl_cmdimpl.c |   14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
tools/libxl/xl_cmdimpl.c +0 -0
index f1c136a..a29a579 100644

tools/libxl/xl_cmdimpl.c int main_vcpupin(int argc, char **argv) · line 4606

Quoted 7 lines
tools/libxl/xl_cmdimpl.c +2 -1
@@ -4601,8 +4601,9 @@ int main_vcpupin(int argc, char **argv)
     libxl_vcpuinfo *vcpuinfo;
     libxl_bitmap cpumap_hard, cpumap_soft;;
     libxl_bitmap *soft = &cpumap_soft, *hard = &cpumap_hard;
-    uint32_t vcpuid, domid;
     const char *vcpu, *hard_str, *soft_str;
+    uint32_t domid;
+    long vcpuid;

I think an int would be sufficiently large for this.

Quoted 5 lines
Quoted 2 lines

The only reason why I used a long is that the function used to do the actual conversion is strtol(), which returns a long.

Can't it use atoi?

From atoi(3):

"The behavior is the same as strtol(nptr, NULL, 10); except that atoi() does not detect errors."

While the current code uses some of the strtol() error handling capabilities:

    /* Figure out with which vCPU we are dealing with */                                                              
    vcpuid = strtoul(vcpu, &endptr, 10);                                                                              
    if (vcpu == endptr) {                                                                                             
        if (strcmp(vcpu, "all")) {                                                                                    
            fprintf(stderr, "Error: Invalid argument.\n");                                                            
            goto out;                                                                                                 
        }                                                                                                             
        vcpuid = -1;                                                                                                  
    }

I can try to reorganize the code a bit but, without being sure that the passed string, if not the string "all", is an actual number (which, AFAIUI, is what strtol() gives me that atoi() doesn't), I don't think we can be as precise in error detection/reporting as we are now (I just double checked, and i=atoi("foo") means i becomes 0).

Quoted 3 lines

I can probably combine atoi() and (something like) isdigit(vcpu[0]) to get to something similar, but at that point, I think I like strtol() better.

Yes, I'd forgotten that this wasn't a straight conversion.

It's probably easier to stick with long instead of farting around range checking the result of strtol.

Ian.

#6 lore raw
In reply to Ian Campbell
Quoted 38 lines · Ian Campbell
Quoted 8 lines · Dario Faggioli
Quoted 7 lines · Ian Campbell
Quoted 3 lines · Dario Faggioli

No functional changes, it just looks more correct, considering that at some point in the function we assign -1 to it (and at some other later point we check for it to be -1),

I think that's more than just "looks more correct", it's just wrong to compare an unsigned to -1, or to assign it.

Quoted 8 lines
Quoted 5 lines

And there may well be a functional change since gcc is no longer allowed to assume that the number is +ve and therefore cannot discard some of these checks any more. We are probably just lucky that gcc isn't doing so already (and unlucky that it isn't generating a warning...)

Right, and thanks for all the details. :-)

Should I change the changelog, if I repost?

Yes, please.

Quoted 19 lines
Quoted 19 lines
Quoted 10 lines
---
 tools/libxl/xl_cmdimpl.c |   14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
tools/libxl/xl_cmdimpl.c +0 -0
index f1c136a..a29a579 100644

tools/libxl/xl_cmdimpl.c int main_vcpupin(int argc, char **argv) · line 4606

Quoted 7 lines
tools/libxl/xl_cmdimpl.c +2 -1
@@ -4601,8 +4601,9 @@ int main_vcpupin(int argc, char **argv)
     libxl_vcpuinfo *vcpuinfo;
     libxl_bitmap cpumap_hard, cpumap_soft;;
     libxl_bitmap *soft = &cpumap_soft, *hard = &cpumap_hard;
-    uint32_t vcpuid, domid;
     const char *vcpu, *hard_str, *soft_str;
+    uint32_t domid;
+    long vcpuid;

I think an int would be sufficiently large for this.

Quoted 3 lines
Quoted 2 lines

The only reason why I used a long is that the function used to do the actual conversion is strtol(), which returns a long.

Can't it use atoi?

No code written this decade should be using atoi() for input sanitisation, given strto*() and friends in C99.

As a prefect example of why, try running xenwatchdogd --help

~Andrew

#7 lore raw
In reply to Ian Campbell
Quoted 38 lines · Ian Campbell
Quoted 8 lines · Dario Faggioli
Quoted 7 lines · Ian Campbell
Quoted 3 lines · Dario Faggioli

No functional changes, it just looks more correct, considering that at some point in the function we assign -1 to it (and at some other later point we check for it to be -1),

I think that's more than just "looks more correct", it's just wrong to compare an unsigned to -1, or to assign it.

Quoted 8 lines
Quoted 5 lines

And there may well be a functional change since gcc is no longer allowed to assume that the number is +ve and therefore cannot discard some of these checks any more. We are probably just lucky that gcc isn't doing so already (and unlucky that it isn't generating a warning...)

Right, and thanks for all the details. :-)

Should I change the changelog, if I repost?

Yes, please.

Quoted 19 lines
Quoted 19 lines
Quoted 10 lines
---
 tools/libxl/xl_cmdimpl.c |   14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
tools/libxl/xl_cmdimpl.c +0 -0
index f1c136a..a29a579 100644

tools/libxl/xl_cmdimpl.c int main_vcpupin(int argc, char **argv) · line 4606

Quoted 7 lines
tools/libxl/xl_cmdimpl.c +2 -1
@@ -4601,8 +4601,9 @@ int main_vcpupin(int argc, char **argv)
     libxl_vcpuinfo *vcpuinfo;
     libxl_bitmap cpumap_hard, cpumap_soft;;
     libxl_bitmap *soft = &cpumap_soft, *hard = &cpumap_hard;
-    uint32_t vcpuid, domid;
     const char *vcpu, *hard_str, *soft_str;
+    uint32_t domid;
+    long vcpuid;

I think an int would be sufficiently large for this.

Quoted 3 lines
Quoted 2 lines

The only reason why I used a long is that the function used to do the actual conversion is strtol(), which returns a long.

Can't it use atoi?

No code written this century should be using atoi() for input conversion/sanitisation, given strto*() and friends in C99.

As a prefect example of why, try running xenwatchdogd --help

~Andrew