Skip to contentSkip to conversation

KVM: TDX: Fix uninitialized error code for __tdx_bringup()

PATCH Superseded Tony Lindgren 3 participants 5 messages Sep 9, 2025

Subsystems (4)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 KVM: TDX: Fix uninitialized error code for __tdx_bringup()

Fix a Smatch static checker warning reported by Dan:

	arch/x86/kvm/vmx/tdx.c:3464 __tdx_bringup()
	warn: missing error code 'r'
Fixes 61bb28279623 ("KVM: TDX: Get system-wide info about TDX module on initialization")
---
 arch/x86/kvm/vmx/tdx.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
arch/x86/kvm/vmx/tdx.c +6 -2
index 66744f5768c8..3ce7fe08afd8 100644
@@ -3466,11 +3466,15 @@ static int __init __tdx_bringup(void)
 
 	/* Check TDX module and KVM capabilities */
 	if (!tdx_get_supported_attrs(&tdx_sysinfo->td_conf) ||
-	    !tdx_get_supported_xfam(&tdx_sysinfo->td_conf))
+	    !tdx_get_supported_xfam(&tdx_sysinfo->td_conf)) {
+		r = -EINVAL;
 		goto get_sysinfo_err;
+	}
 
-	if (!(tdx_sysinfo->features.tdx_features0 & MD_FIELD_ID_FEATURES0_TOPOLOGY_ENUM))
+	if (!(tdx_sysinfo->features.tdx_features0 & MD_FIELD_ID_FEATURES0_TOPOLOGY_ENUM)) {
+		r = -EINVAL;
 		goto get_sysinfo_err;
+	}
 
 	/*
 	 * TDX has its own limit of maximum vCPUs it can support for all
Signature · 1 line
2.43.0
#1 lore raw
Quoted 24 lines · Tony Lindgren

Fix a Smatch static checker warning reported by Dan:

	arch/x86/kvm/vmx/tdx.c:3464 __tdx_bringup()
	warn: missing error code 'r'
Fixes 61bb28279623 ("KVM: TDX: Get system-wide info about TDX module on initialization")
---
 arch/x86/kvm/vmx/tdx.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
arch/x86/kvm/vmx/tdx.c +2 -1
index 66744f5768c8..3ce7fe08afd8 100644
@@ -3466,11 +3466,15 @@ static int __init __tdx_bringup(void)
 
 	/* Check TDX module and KVM capabilities */
 	if (!tdx_get_supported_attrs(&tdx_sysinfo->td_conf) ||
-	    !tdx_get_supported_xfam(&tdx_sysinfo->td_conf))
+	    !tdx_get_supported_xfam(&tdx_sysinfo->td_conf)) {
+		r = -EINVAL;
 		goto get_sysinfo_err;

arch/x86/kvm/vmx/tdx.c static int __init __tdx_bringup(void) · line 3480

Quoted 10 lines
arch/x86/kvm/vmx/tdx.c +4 -1
+	}
 
-	if (!(tdx_sysinfo->features.tdx_features0 & MD_FIELD_ID_FEATURES0_TOPOLOGY_ENUM))
+	if (!(tdx_sysinfo->features.tdx_features0 & MD_FIELD_ID_FEATURES0_TOPOLOGY_ENUM)) {
+		r = -EINVAL;
 		goto get_sysinfo_err;
+	}
 
 	/*
 	 * TDX has its own limit of maximum vCPUs it can support for all

Looking at the code, seems all checks after tdx_get_sysinfo() have the same pattern when the check fails:

	r = -EINVAL;
	goto get_sysinfo_err;

How about we just initialize r to -EINVAL once before tdx_get_sysinfo() so that all 'r = -EINVAL;' can be removed? I think in this way the code would be simpler (see below diff [*])?

The "Fixes" tag would be hard to identify, though, because the diff touches the code introduced multiple commits. But I am not sure whether this is a true issue since AFAICT we can use multiple "Fixes" tags.

[*] the diff:

arch/x86/kvm/vmx/tdx.c +2 -3
index 7b81cd1fbba5..113cb074b90c 100644
@@ -3460,12 +3460,11 @@ static int __init __tdx_bringup(void)
        if (r)
                goto tdx_bringup_err;
 
+       r = -EINVAL;
        /* Get TDX global information for later use */
        tdx_sysinfo = tdx_get_sysinfo();
-       if (WARN_ON_ONCE(!tdx_sysinfo)) {
-               r = -EINVAL;
+       if (WARN_ON_ONCE(!tdx_sysinfo))
                goto get_sysinfo_err;
-       }
 
        /* Check TDX module and KVM capabilities */
        if (!tdx_get_supported_attrs(&tdx_sysinfo->td_conf) ||
@@ -3508,14 +3507,11 @@ static int __init __tdx_bringup(void)
        if (td_conf->max_vcpus_per_td < num_present_cpus()) {
                pr_err("Disable TDX: MAX_VCPU_PER_TD (%u) smaller than

number of logical CPUs (%u).\n",

                                td_conf->max_vcpus_per_td,
num_present_cpus());
-               r = -EINVAL;
                goto get_sysinfo_err;
        }

- if (misc_cg_set_capacity(MISC_CG_RES_TDX,
tdx_get_nr_guest_keyids())) {
- r = -EINVAL;
+ if (misc_cg_set_capacity(MISC_CG_RES_TDX,
tdx_get_nr_guest_keyids()))

                goto get_sysinfo_err;

- }

/*
* Leave hardware virtualization enabled after TDX is enabled

#2 lore raw
Quoted 7 lines · Kai

How about we just initialize r to -EINVAL once before tdx_get_sysinfo() so that all 'r = -EINVAL;' can be removed? I think in this way the code would be simpler (see below diff [*])?

The "Fixes" tag would be hard to identify, though, because the diff touches the code introduced multiple commits. But I am not sure whether this is a true issue since AFAICT we can use multiple "Fixes" tags.

Your diff looks fine to me, however my personal preference would be to do the fix first then clean-up :)

Regards,

Tony

#3 lore raw
Quoted 6 lines · Tony Lindgren
Quoted 5 lines · Kai

How about we just initialize r to -EINVAL once before tdx_get_sysinfo() so that all 'r = -EINVAL;' can be removed? I think in this way the code would be simpler (see below diff [*])?

The "Fixes" tag would be hard to identify,

No, Fixes always points at the commit(s) that introduced buggy behavior. While one might argue that commit 61bb28279623 was set up to fail by earlier commits, that commit is unequivocally the one and only Fixes commit.

Quoted 6 lines
Quoted 3 lines

though, because the diff
touches the code introduced multiple commits. But I am not sure whether this is a true issue since AFAICT we can use multiple "Fixes" tags.

Your diff looks fine to me, however my personal preference would be to do the fix first then clean-up :)

Eh, fixes can also harden against similar failures in the future. I don't see any reason to split this one up. The buggy commit was introduced in v6.16 and Kai's suggestion applies cleanly there, so the more aggressive fix won't lead to stable@ conflicts either.

In short, let's go straight to Kai's version.

#4 lore raw
Quoted 11 lines · Sean Christopherson
Quoted 6 lines · Tony Lindgren
Quoted 5 lines · Kai

How about we just initialize r to -EINVAL once before tdx_get_sysinfo() so that all 'r = -EINVAL;' can be removed? I think in this way the code would be simpler (see below diff [*])?

The "Fixes" tag would be hard to identify,

No, Fixes always points at the commit(s) that introduced buggy behavior. While one might argue that commit 61bb28279623 was set up to fail by earlier commits, that commit is unequivocally the one and only Fixes commit.

Good to know. Thanks Sean!