linux-kernel - Re: [PATCH v2] PCI: pci-bridge-emul: Set position of PCI capabilities to real HW value

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Wed, 24 Aug 2022 12:33:08 +0200
From:   Pali Rohár <pali@...nel.org>
To:     Lorenzo Pieralisi <lpieralisi@...nel.org>
Cc:     Bjorn Helgaas <bhelgaas@...gle.com>,
        Krzysztof Wilczyński <kw@...ux.com>,
        Rob Herring <robh@...nel.org>,
        Thomas Petazzoni <thomas.petazzoni@...tlin.com>,
        Marek Behún <kabel@...nel.org>,
        linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] PCI: pci-bridge-emul: Set position of PCI
 capabilities to real HW value

On Wednesday 24 August 2022 09:25:15 Lorenzo Pieralisi wrote:
> [...]
> 
> > > In other words: what if PCI express capability is lower in the address
> > > space (config) than the subsystem ID vendor capability ?
> > 
> > Current code expects that if host controller driver sets both pcie_start
> > and ssid_start, those values are correct, non-overlapping and can be
> > handled correctly.
> > 
> > And if offset to PCI express capability is lower than offset to SSID
> > capability then there should not be any issue. First capability is
> > correctly set into capabilities_pointer (via min function) and then
> > pci_bridge_emul_conf_read() should handle it.
> 
> I don't understand how the pointer to the SSID cap is set in this
> specific case, I don't see any code doing that but that's most certainly
> because I don't know in details the emul bridge internals.
> 
> IIUC, in pci_bridge_emul_read_ssid(), we set the next cap pointer if
> the PCI express capability is at an address higher (current case) than
> the SSID capability otherwise we set it to 0 (end of the list).
> 
> The other way around, I don't see the PCI express next cap pointer
> being set anywhere (where, IIUC, it should be set to point to the
> SSID cap) - I am not sure you are handling it.

Ou, I see. This is really missing! It should be set when initializing
pcie_conf.cap_id and pcie_conf.cap at this place:
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/drivers/pci/pci-bridge-emul.c?h=v6.0-rc2#n379

I will fix it and send a new version of this patch.

> That's the only question I have on this patch.
> 
> Lorenzo
> 
> > The whole my idea is to construct capabilities linked list structure
> > correctly based on input requirements (e.g. fixed location of some
> > capability, etc).
> > 
> > > I am just trying to understand the patch, so forgive me if the question
> > > is already addressed in the code.
> > > 
> > > Thanks,
> > > Lorenzo
> > > 
> > > > +		else
> > > > +			bridge->pcie_start = bridge->ssid_start + PCI_CAP_SSID_SIZEOF;
> > > > +	}
> > > > +
> > > > +	bridge->conf.capabilities_pointer = min(bridge->ssid_start, bridge->pcie_start);
> > > >  
> > > >  	if (bridge->conf.capabilities_pointer)
> > > >  		bridge->conf.status |= cpu_to_le16(PCI_STATUS_CAP_LIST);
> > > > @@ -459,15 +468,17 @@ int pci_bridge_emul_conf_read(struct pci_bridge_emul *bridge, int where,
> > > >  		read_op = bridge->ops->read_base;
> > > >  		cfgspace = (__le32 *) &bridge->conf;
> > > >  		behavior = bridge->pci_regs_behavior;
> > > > -	} else if (reg >= PCI_CAP_SSID_START && reg < PCI_CAP_SSID_END && bridge->subsystem_vendor_id) {
> > > > +	} else if (reg >= bridge->ssid_start && reg < bridge->ssid_start + PCI_CAP_SSID_SIZEOF &&
> > > > +		   bridge->subsystem_vendor_id) {
> > > >  		/* Emulated PCI Bridge Subsystem Vendor ID capability */
> > > > -		reg -= PCI_CAP_SSID_START;
> > > > +		reg -= bridge->ssid_start;
> > > >  		read_op = pci_bridge_emul_read_ssid;
> > > >  		cfgspace = NULL;
> > > >  		behavior = NULL;
> > > > -	} else if (reg >= PCI_CAP_PCIE_START && reg < PCI_CAP_PCIE_END && bridge->has_pcie) {
> > > > +	} else if (reg >= bridge->pcie_start && reg < bridge->pcie_start + PCI_CAP_PCIE_SIZEOF &&
> > > > +		   bridge->has_pcie) {
> > > >  		/* Our emulated PCIe capability */
> > > > -		reg -= PCI_CAP_PCIE_START;
> > > > +		reg -= bridge->pcie_start;
> > > >  		read_op = bridge->ops->read_pcie;
> > > >  		cfgspace = (__le32 *) &bridge->pcie_conf;
> > > >  		behavior = bridge->pcie_cap_regs_behavior;
> > > > @@ -538,9 +549,10 @@ int pci_bridge_emul_conf_write(struct pci_bridge_emul *bridge, int where,
> > > >  		write_op = bridge->ops->write_base;
> > > >  		cfgspace = (__le32 *) &bridge->conf;
> > > >  		behavior = bridge->pci_regs_behavior;
> > > > -	} else if (reg >= PCI_CAP_PCIE_START && reg < PCI_CAP_PCIE_END && bridge->has_pcie) {
> > > > +	} else if (reg >= bridge->pcie_start && reg < bridge->pcie_start + PCI_CAP_PCIE_SIZEOF &&
> > > > +		   bridge->has_pcie) {
> > > >  		/* Our emulated PCIe capability */
> > > > -		reg -= PCI_CAP_PCIE_START;
> > > > +		reg -= bridge->pcie_start;
> > > >  		write_op = bridge->ops->write_pcie;
> > > >  		cfgspace = (__le32 *) &bridge->pcie_conf;
> > > >  		behavior = bridge->pcie_cap_regs_behavior;
> > > > diff --git a/drivers/pci/pci-bridge-emul.h b/drivers/pci/pci-bridge-emul.h
> > > > index 71392b67471d..2a0e59c7f0d9 100644
> > > > --- a/drivers/pci/pci-bridge-emul.h
> > > > +++ b/drivers/pci/pci-bridge-emul.h
> > > > @@ -131,6 +131,8 @@ struct pci_bridge_emul {
> > > >  	struct pci_bridge_reg_behavior *pci_regs_behavior;
> > > >  	struct pci_bridge_reg_behavior *pcie_cap_regs_behavior;
> > > >  	void *data;
> > > > +	u8 pcie_start;
> > > > +	u8 ssid_start;
> > > >  	bool has_pcie;
> > > >  	u16 subsystem_vendor_id;
> > > >  	u16 subsystem_id;
> > > > -- 
> > > > 2.20.1
> > > >