Repository navigation
fullhan: FH885xV200, FH865x, FH885x V100, FH8626V100 and FH8632 machines - #207
Merged
Merged
Conversation
Ten machines in hw/arm/fullhan.c over one memory map of mostly DesignWare blocks, with Fullhan's own devices as hw/*/fh-*.c: * fh-intc (DW APB ICTL; byte reads of FINALSTATUS for find_first_bit), fh-timer (DW APB timers clocked from the PMU), fh-uart (16550 + USR), fh-gpio, fh-spi (DW SSI with the SPIC data window, swap and 32-bit frames; fast-read dummies clocked as 8), fh-axi-dmac / fh-ahb-dmac, fh-i2c, fh-wdt, fh-rtc (v2 command interface, FH8632's v1), fh-pmu (chip IDs, boot strap, resets, the timer clock as a QEMU Clock, PTS), fh-efuse, fh-pts, fh-aes (AES/DES/3DES in ECB/CBC/CTR/CFB/OFB through QEMU's ciphers) and fh-sadc (the three register layouts); * the GMAC on xm-dwmac (new phy-id / phy-addr properties), MMC on hisi-himci, USB on dwc2-usb, the FH865x GIC-400 on a15mpcore_priv; * a mask ROM stand-in that loads the boot header's image, past FH865x's DDR-sizing ramboot; the SPI NOR from flash-file= (written back); * sensor=: a GalaxyCore model (GC4653, GC2053, GC2063, GC2093, GC1054) or the MIS2006 / JX-F37 ones on I2C0, for the vendor apps and ipcinfo. QEMU patches: 0012 (CCSIDR of a missing cache reads the L1 D-cache's, for FH865x's U-Boot), 0013 (m25p80 clears WEL after program / erase, for the 3.0.8 kernels' m25p80), 0014 (dwc2 GHWCFG properties and host descriptor DMA, for the 3.0.8 kernels' fh_otg). Camera dumps boot through the vendor U-Boot to their applications; the OpenIPC fh8852v200 / fh8852v100 releases boot to a login with a DHCP lease and a JFFS2 overlay (qemu-boot/test-fullhan-boot.sh, in CI). The ARC coprocessor and the media blocks are not modelled.
PR Summary by QodoAdd ten Fullhan camera SoC machines and firmware boot support
AI Description
Diagram
High-Level Assessment
Files changed (30)
|
Code Review by Qodo
1.
|
![qodo-free-for-open-source-projects[bot]](https://avatars.githubusercontent.com/in/1068000?s=60&v=4){kind=small}
Comment on lines
+170
to
+173
| if (s->xfer != want || (cmd & CMD_RESTART)) { | ||
| s->raw |= INTR_START_DET | INTR_ACTIVITY; | ||
| if (i2c_start_transfer(s->bus, addr, rd)) { | ||
| fh_i2c_abort(s, ABRT_7B_ADDR_NOACK); |
There was a problem hiding this comment.
4. Repeated-start reads retain the old transfer 🐞 Bug ≡ Correctness
fh_i2c_data_cmd() calls i2c_start_transfer() for a repeated start without ending its active QEMU I²C transfer. A write-then-repeated-start-read sequence therefore skips the target's transfer-end handling; the other I²C controller explicitly performs that transition before starting again.
Agent Prompt
## Issue description
Repeated-start commands start another QEMU I²C transfer without closing the active one.
## Fix Focus Areas
- qemu/hw/i2c/fh-i2c.c[161-177]
## Recommended Fix
End an active bus transfer before calling `i2c_start_transfer()` for a direction change or explicit restart, and keep the controller's transfer state consistent on failure.
ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools
Collaborator
Author
There was a problem hiding this comment.
Partly: QEMU's I2C core treats a start on a busy bus as a repeated start by design (no END in between, matching I2C, where there is no STOP), so that is intended. The real gap was a repeated start to a different address keeping the old target; fixed in 3de1e4e by ending the first transfer in that case.
…reads * fh-axi-dmac / fh-ahb-dmac: a failed read or write stops the channel with SRC_DEC_ERR / DST_DEC_ERR (AXI) or ERR (AHB) instead of carrying on to transfer-complete; every loop iteration spends the yield budget, so a cycle of zero-length LLIs no longer pins the main loop. * fh-spi: in EEPROM mode the response is clocked in once the command phase is over (the next register read, or 10 us after the last command frame), not between the frames of a multi-frame command. * fh-i2c: a repeated start to another address ends the transfer with the first target before addressing the second. * fh-aes: transfers move through a fixed 4 KiB buffer, not one sized by the guest's DMA length. * dwc2 (patch 0014): isochronous channels under descriptor DMA are refused (LOG_UNIMP, XactErr) rather than read with the non-isochronous descriptor layout. * test-fullhan.py: every machine's chip / package IDs and a migration round trip for all ten.
…ckend CI builds QEMU without gnutls, nettle or gcrypt, and the stub backend refuses every cipher: the engine moved no data and test-fullhan.py's AES checks read back zeros. AES now uses crypto/aes.h, which QEMU always builds; DES and 3DES still need a backend and are logged as unimplemented without one, and the qtest skips the DES check when the build has none.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds Fullhan's camera SoCs: ten machines in
hw/arm/fullhan.cover one memory map of mostly DesignWare blocks, Fullhan's own devices ashw/*/fh-*.c. Details indocs/fullhan.md.Machines:
fh8852v200,fh8856v200,fh8858v200,fh8652,fh8656,fh8658,fh8852v100,fh8856v100,fh8626v100,fh8632.Devices
qom-set /machine/sadc chN).xm-dwmac(newphy-id/phy-addrproperties), MMC onhisi-himci, USB ondwc2-usb, the FH865x GIC-400.sensor=on I2C0: GalaxyCore GC4653 / GC2053 / GC2063 / GC2093 / GC1054, MIS2006, JX-F37.flash-file=, written back), past FH865x's DDR-sizing ramboot.QEMU patches
fh_otg).Tested
fh8852v200/fh8852v100boot to a login with a DHCP lease and a JFFS2 overlay:qemu-boot/test-fullhan-boot.sh, added to CI.qemu/tests/test-fullhan.py(qtest, all generations; AES against FIPS-197 / SP 800-38A vectors), added to CI.usb-netpings overfh_otg.Not modelled: the ARC625 coprocessor and the ISP / VPU / encoder / JPEG blocks (only binaries in the SDKs), hence no video or audio; PWM, IR receiver.